EPA-460/3-74-001
DIESEL FUEL INJECTION
SYSTEM
SIMULATION AND EXPERIMENTAL
CORRELATION
I .S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Water Programs
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
-------
EPA-460/3-74-001
DIESEL FUEL INJECTION
SYSTEM
SIMULATION AND EXPERIMENTAL
CORRELATION
Prepared by
J. A. Bolt, M. F. El-Erian, and E. B. Wylie
Automotive Engineering Laboratory
Department of Mechanical Engineering
College of Engineering
The University of Michigan
Ann Arbor, Michigan 48105
Grant No. R800 424
EPA Project Officer:
Dr. Jose L. Bascunana
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Water Programs
Office of Mobile Source Air Pollution Control
Emission Control Technology Division
Ann Arbor, Michigan 48105
Januany 1974
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors
and grantees, and nonprofit organizations - as supplies permit - from
the Air Pollution Technical Information Center, Environmental Protec-
tion Agency, Research Triangle Park5 North Carolina 27711, or from the
National Technical Information Service, 5285 Port Royal Road, Spring-
field, Virginia 22151.
This report was furnished to the Environmental Protection Agency by the
Automotive Engineering Laboratory, Ann Arbor, Michigan, in fulfillment of
Grant No. R800 424. The contents of this report are reproduced herein as
received from the Automotive Engineering Laboratory. The opinions, findings,
and conclusions expressed are those of the author and not necessarily those
of the Environmental Protection Agency. Mention of company or product names
is not to be considered as an endorsement by the Environmental Protection Agency.
Publication No. EPA-460/3-74-001
n
-------
ABSTRACT
After-injection is the term used to describe the introduction of additional
fuel into the combustion chamber of a diesel engine after the end of the normal
injection. After-injection is caused by uncontrolled pressure transients in
the injection system after the opening of the pump spill port. These pressure
transients are related to the wave propagation phenomena in the high-pressure
pipeline connecting the pump and injector. It is a persistent diesel fuel in-
jection system problem which results in reduced engine power and economy and
increased emissions.
A theoretical digital simulation of a conventional diesel fuel injection
system was developed. The influence of such factors as wave progagation phe-
nomena, pipe friction, and cavitation are included. The computer results are
compared with transient pressures as measured on an actual fuel injection system
operated on a test bench. The comparisons show the accuracy and validity of
this simulation method. Special attention is given to some of the important
factors that affect the accuracy of the simulation model. These include the
effect of pressure on the fuel bulk modulus and wave speed, the pipeline re-
sidual pressure, and the coefficient of discharge of important orifices.
Analytical control methods were developed to help determine design means
by which after-injection may be controlled. Further Investigation and evalua-
tion of two design changes which release the injection system excess elastic
energy in a controlled manner are considered in the report. One design change
consists of the addition of a control valve in the pump delivery chamber. The
other involves the modification of the pump spill port. In both cases, pres-
sures and flows are not significantly altered during the main injection period.
The ability of both design changes to control after-injection is confirmed by
the use of a simulation program. Experimental data from a system with the
pump spill port modified in accordance with theoretical design calculations
provided satisfactory confirmation of the analyses.
iii
-------
SUMMARY
Difficult problems exist in diesel fuel injection systems due to pressure
waves and other highly transient phenomena in the high pressure lines and
other components. Among other problems, these transient pressure phenomena
commonly give rise to a condition known as after-injection, the name used to
denote a condition when the fuel spray nozzle discharges an additional quantity
of fuel after the normal fuel delivery. After-injection is mainly the result
of pressure wave phenomena traveling the length of the line connecting the
pump and nozzle. It can also result from other sources of elasticity in the
complete system.
We believed, based on experience, and from talking to many engineers very
familiar with diesel engines, that this condition of after-injection was much
more prevalent and a greater handicap to good engine performance than has been
indicated by the technical literature concerning diesel engines. In recent
years there has been conjecture concerning the effect of after-injection on
diesel engine exhaust emissions and smoke.
Interest in this project concerning diesel fuel injection system simula-
tion also arose in part from our research work done for the U.S. Army Tank
Automotive Command (TACOM). This work for the Army under contracts* during
the past eight years was concerned with ignition delay and combustion phenomena
under conditions of high supercharging pressure, and with high cylinder coolant
temperatures. In the course of this work we first used a single-cylinder
Lister-Blackstone diesel engine, and later the U.S. Army Tank Command (TACOM)
engine built by the International Harvester Company. The TACOM engine had
substantial amounts of after-injection at many operating conditions which
seriously affect the reproducibility of data, the engine performance, and
exhaust emissions. These technical difficulties are described in a final
Army report listed below.**
A much more complete understanding of the transient phenomena in the jerk-
pump injection system (for example, as made by American Bosch) was gained by
developing a digital computer simulation program. The results of this program
were checked experimentally by means of a one-cylinder fuel injection system
tested on a bench test rig. This system was the same as that used on the TACOM
research engine.
^-Contract Numbers DAAE07-71-C-02148 and DAAE07-69-1289-
**U-S. Army Tank Automotive Command, Technical Report No. 11796, entitled,
"Diesel Engine Combustion and Emissions—TACOM Research Engine/' by Bolt,
El-Erian, Duerr, and Milkos, April, 1975.
-------
The analytical studies and digital simulation work was greatly aided by
enlisting the help of Professors E. Benjamin Wylie and Victor L. Streeter of
The University of Michigan, Department of Civil Engineering, who are well
known for their simulation of transient phenomena in large hydraulic power
systems. They, with our graduate students,, applied the method of character-
istics to the relatively small diesel systems. As a result of this work, one
doctoral thesis was prepared, and three SAE papers were published. These de-
scribe in detail the technical work done and the results accomplished. These
are as follows:
1. "Diesel Fuel Injection System Simulation and Experimental Correlation,
by E. Benjamin Wylie, Jay A. Bolt, and Mohamed F. El-Erian, Society
of Automotive Engineers, Paper No. 710569, June, 1971, SAE Trans.^
Vol. 80, 1971.
2. "Simulation and Control of Transient Flow in the Diesel Injection
System," Mohamed F. El-Erian. A University of Michigan doctoral
thesis, published in 1972.
3. "Analysis and Control of Transient Flow in the Diesel Injection
System, Part I - The Analytical Control Method," by Mohamed F.
El-Erian, E. Benjamin Wylie, and Jay A. Bolt, Society of Automotive
Engineers, Paper Wo. 73066l, June, 1975, SAE Trans., Vol. 82, 1975,
U. "Analysis and Control of Transient Flow in the Diesel Injection
System, Part II - Design Results of Controlled After-Injection," by
Mohamed F. El-Erian, E. Benjamin Wylie, and Jay A. Bolt, Society of
Automotive Engineers, Paper No. 750662, June, 1973; SAE Trans., Vol.
82, 1975.
The body of this report consists of the three papers listed above, which
are attached at the end of this report. The doctoral thesis of M. F. El-Erian,
item 2 above, also contains much detailed information concerning the simula-
tion work. Copies of this University of Michigan thesis, consisting of 151
pages, are available from University Microfilms, 300 N. Zeeb Road, Ann Arbor,
Michigan, for approximately $15.00
Under the grant we had hoped to install the standard and modified injec-
tion systems on the TACOM research diesel engine. This would make it possible
to operate the injection system and engine with or without after-injection,
with all other conditions being held constant. This would permit us to obtain
accurate quantitative data on the effect of after-injection on diesel engine
performance, emissions, and smoke. However, unforeseen difficulties with the
test equipment made it impractical to complete this phase of the project be-
fore the grant funds were exhausted.
We are, however, now doing this comparative testing with very limited
research funds available from several small industry research grants provided
to the Department of Mechanical Engineering for graduate student support.
-------
With this support a doctoral student from Nigeria, Mr, Sam Onyegegbu, is pro-
ceeding with this comparative testing as the topic of his doctoral thesis.
We are hopeful that will will obtain reliable data suitable for publica-
tion showing the effect of known amounts and characteristics of after-injection
on the performance, emissions, and smoke from diesel engines.
ACKNOWLEDGMENT
This investigation was supported by the United States Environmental
Protection Agency, as indicated on the cover sheet. The cooperation of the
following persons is greatly appreciated: Dr. Jose Bascunana of the EPA;
Mr. Floyd Lux, George Cheklich, and others of the U.S. Army Tank-Automotive
Command. The valuable contributions of doctoral students Bela Petry and Sam
Onyegegbu are also appreciated.
-------
SOCIETY OF AUTOMOTIVE ENGINEERS, INC
Two Pennsylvania Plaza, New York, N.Y. 10001
Diesel Fuel Injection System
Simulation and Experimental
Correlation
E. Benjamin Wylie, Jay A. Bolt, and Mohamed F. EI-Erian
University of Michigan
SOCIETY OF AUTOMOTIVE ENGINEERS
Mid-Year Meeting
Montreal, due., Can.
June 7-11,1971
710569
-------
E. Benjamin Wylie, Jay A. Bolt, and Mohamed F. Ef-Erian
University of Michigan
THE PURPOSE OF THIS PAPER is to report transient phe-
nomena measured in a diesel fuel injection system and to de-
scribe an analytical simulation of the system. Comparisons
are made between experimental and computed data for sev-
eral different operating conditions of the injection system.
The characteristics unique to this simulation are discussed in
some detail.
A diesel fuel injection system is an assembly of many com-
plex and intricate mechanical components, each with a spe-
cific function. In addition to the behavior of individual com-
ponents, the interaction of these components has an impor-
tant influence on the ultimate operation of the system. The
response of the fuel injector is primarily dependent upon the
action of the pump and the pressure wave propagation phe-
nomena in the delivery pipeline. The pump action, with its
delivery chamber and valve, is not independent of the injec-
tor and delivery pipeline; that is, a complete system interac-
tion takes place.
An understanding of the detailed operation of individual
components in the system and the influence of design param-
eters on this operation are the desired objectives. With such
information available in an analytical simulation, perhaps
design changes may be identified to improve nozzle fuel
spray characteristics, to eliminate the secondary after-injec-
tion, and to generally improve the engine combustion char-
acteristics with a view towards reduced smoke and exhaust
emissions.
In this paper, following a brief review of the topical litera-
ture, the analytical model and experimental apparatus are de-
scribed. Comparisons between simulated and experimental re-
sults are then presented, followed by a critical review of the
results. Significant parameters and control variables are iden-
tified and discussed, and the usefulness of the analytical simu-
lation is intimated.
LITERATURE REVIEW
A great variety of procedures have been used to study the
diesel injection system, some experimental and others theoret-
ical in nature. Before 1960, the injection simulation studies
were limited due to the requirement of lengthy mathematical
computations. This disadvantage made it impractical to apply
theoretical simulation in the design stage and led to experi-
mental tnai-and-error procedures. Analytical studies were
completed only after the introduction of many simplifying
assumptions.
• ABSTRACT•
A theoretical digital simulation of a conventional diesel fuel
injection system has been developed. The influence of such
factors as wave propagation phenomena, pipe friction, and
cavitation are included. The computer results are compared
with transient pressures as measured on an actual fue! injection
system operated on a test bencii. The comparisons show the
accuracy and validity of this simulation scheme. Special atten-
tion is given to some of the important factors that affect the
accuracy of the simulation model. These include the effect of
pressure on the fuel bulk modulus and wave speed, the pipe
line residual pressure, and the coefficient of discharge of im-
portant orifices.
7
-------
One of the earliest significant contributions relative to fuel
injection systems is due to Davis and Giffen (1)*. Their
discussion includes mention of many of the significant vari-
ables involved in the system: fluid compressibility, elastic de-
formation, pressure wave propagation, fluid friction, and
pump and nozzle characteristics, including secondary injection.
De Jtihasz(2),used graphical water hammer concepts to pro-
vide an analysis of a linear model of typical simplified injection
systems, including the elements of the pipeline, pump, nozzle,
and a fluid volume.
Giffen and Row (3), theoretically solved the equations rep-
resenting the injection system, taking into account the effect
of pressure waves in the delivery pipe and the capacity effects
of the volumes concentrated in the pump and nozzle system.
They handled the differential equations by placing them in fi-
nite difference form and finding an algebraic expression for
the solution. This method of solution was limited to simple
injection models because of the time required for mathemati-
cal solutions.
Knight (4), introduced a model for viscous friction and cavi-
tation in the delivery pipe, and used the same model for the
pump and nozzle system described in Rcf. 3. His calculations
were performed on a digital computer.
Becchi (5), used a model which comprised a detailed repre-
sentation of the injector and the pump, but he neglected
friction in the delivery pipe and had no provision for possible
occurrence of vapor cavities. He solved the system of differen-
tial equations by an iterative method after writing them in fi-
nite difference form.
Brown and McCallion (6), combined Becchi's detailed repre-
sentation of the pump and injector with a model that included
viscous friction and possible cuvitation in the delivery pipe.
They also considered a detailed modeling of the delivery valve
as described by Stone (7), and solved the system of equations
by another iterative method.
The work of Walwijk, Van der Graaf, and Jansen (8) is also
to be noted. Their experimental apparatus enabled them actu-
ally to measure the motion of the delivery valve and injector
needle, as well as the pressure in various locations in the sys-
tem. Particular attention was devoted to the motion of the de-
livery valve in their simulation on a digital computer. A good
correlation was achieved between experimental and computed
results.
All of the investigators discuss some of the factors that are
likely to affect the accuracy of the model. The value of the
delivery pipeline base pressure is important for a meaningful
comparison, between the model and experimental results. A
treatment of vapor pressure in the delivery pipeline is also
needed for a complete model. Kreith and Eisentadt (9), and
Lichtarowicv, Duggins, and Markland (10), presented experi-
mental results of the variation of the coefficient of discharge
over a wide range of Reynolds number and length-to-diamcter
ratio. Giffen and Row (3) cautioned of the danger of using co-
efficients of discharge from the literature. They preferred to
*Numbers in parentheses designate References at end of
paper.
use experimentally determined values for the particular nozzle
under consideration. The data of Gelalles (! 1), in which he
tested different nozzle configurations, showed that the coeffi-
cient of discharge, besides depending on iength-over-diameter
ratio and Reynold's number, is also greatly dependent on the
configuration of the reservoir leading to the nozzle holes.
Recent investigators also give considerable attention to the
stability and convergence of their solutions. Hennci (12), dis-
cusses three different methods of numerical solution of a sys-
tem of differential equations: the iterative solution of simulta-
neous algebraic equations, the expansion methods (Taylor's
method or Runge Kutta method), and the numerical integra-
tion methods. The third method includes the predictor-
corrector method which offers the advantage of an adjustable
time increment, dependent upon a given error bound. This
particular advantage is of great value especially for reducing
computation time. The first and second methods require the
use of very small time steps and a prior knowledge of the size
of the time step.
ANALYTICAL MODEL
A schematic representation of the system is shown in Fig. I
and includes three major components: the fuel pump, the con-
necting fuel line, and the fuel injector. The analytical model,
which is programmed for a time variant simulation on the dig-
ital computer, must include an accurate description of the geo-
metric and physical character of the system, as well as the
equations that describe the dynamics of the fluid and mechan-
ical components. A total system analysis is necessary wherein
the ordinary differential equations which describe fluid com-
pressibility, delivery valve motion, and injector needle motion,
are handled numerically simultaneously with the numerical so-
lution of the partial differential equations that describe the
wave propagation phenomena in the fuel lines.
Prior to the description of the formulation of equations and
the method of solution, the underlying assumptions in this
simulation are set forth. Fluid compressibility is introduced
by use of the bulk modulus of elasticity. The variation of bulk
modulus, and the variation of fluid density, with pressure is
also considered (13, 14). All elastic deformations of solid
parts of the system due to pressure changes are neglected. The
foregoing assumptions yield a pressure-dependent wave propa-
gation velocity that is a function of fluid compressibility only.
The error in wave velocity as a result of neglecting pipeline de-
formation is less than 0.5% for the maximum pressure varia-
tion. A distributed parameter one-dimensional model is used
to describe flow in the fuel lines. This implies that the param-
eters of compliance, inertia, and fnctional losses are distrib-
uted along the pipeline. Frictional effects are evaluated by
considering the loss during unsteady flow to be the same as the
loss for steady flow at the same velocity and fluid property. A
friction-factor-resistance formulation is used wherein the fac-
tor is a function of Reynolds number. If vapor pressure is
reached at any point in the system, a vapor cavity is permitted
to grow and collapse in accordance with the dynamic equa-
tions and a local mass continuity balance. Orifice discharge
-------
Fig. 1 - Schematic representation of diesel injection system
and points of pressure measurement
DELIVERY CHAMBER
TRANSDUCER *
NOZZLE UPPER
CHAMBER
NOZZLE LOWER
CHAMBER
TRANSDUCER *5
INJECTION CHAMBER
PUMPING CHAMBER
SUPPLY PORT^X
STRAIN
GAUGE
TRANSDUCER
FEED CHAMBER
TRANSDUCER
'•l
RELIEF VALVE
SPILL PORT
coefficients are based upon steady-state data and may be
Reynolds number dependent.
THEORETICAL FORMULATION OF MODEL - The equa-
tions to describe the dynamic response of the system are
grouped into three categories: the pipeline, the pump, and the
injector.
Transient behavior of a compressible fluid in a pipeline may
be described by the general equations of motion and continu-
ity applied to an elemental length of the pipe. The equation
of motion is expressed in the following form (15):
dt
gA 9p
7 9x
f
2DA
QIQI = 0
(1)
where:
P
Q
A
&
D
f
7
xand t
Pressure
Volumetric flow rate
Pipe cross-sectional area
Acceleration of gravity
Pipe diameter
Darcy-Weisbach friction factor
Unit weight of fluid
Independent variables distance and time.
The absolute value sign is used on the friction term so the
equation is valid for negative flow. The continuity equation
for a horizontal pipeline is:
30, gA_ 3p
9x .2^ 3t
(2)
where:
a = Wave propagation velocity in fluid, defined by
a = Vg K/7.
K = Bulk modulus of elasticity.
The convective acceleration terms in Eqs. 1 and 2 have been
dropped as, in general, they are of much smaller order than
the terms shown.
Eqs. 1 and 2 are transformed, by use of the method of
characteristics (15) into two ordinary differential equations,
each of which is valid along a particular characteristic line.
When these equations are placed in finite-difference form by a
first-order approximation of the integration, they become:
Qz - Qw
gA
(pz - pw)
fwAtQwlQwl
2DA
= 0
X "
XW =
(3)
(4)
QZ - QY -
gA
7YaY
(PZ -
fy AtQy iQyl
2DA
= 0
x^ - Xy = - ay At
where:
subscripts = Positions in x-t plane as shown in Fig. 2.
(5)
(6)
-------
t-
i
t0»3At
t0+2At
T
i
-AX—«
Ungth X
Fig. 2 - Characteristics in x-t plane
Eq. 3 'is valid only along the forward characteristic line, W-Z,
described by Eq. 4, and Eq. 5 is valid only along a receding
characteristic line, Y-Z, described by Eq. 6. An analysis of any
pipeline begins with known conditions at time, t = IQ', then cal-
culations of transient pressures and discharges are made at
fixed, equally spaced sections throughout the pipe. For stabil-
ity reasons it is necessary that the grid spacing, Ax, must be
greater than or equal to the product of the wave speed a, and
At. In Fig. 2, assume p and Q are known at each grid intersec-
tion along the axis t = tQ. At point Z, new values of p and Q
may be calculated by use of Eqs. 3 and 5. In order to accom-
plish this solution, it is necessary to interpolate between
known conditions at A and C to find the values at W. This in-
terpolation is necessary since the variable wave speed may po-
sition W at any location between A and C. A similar inter-
polation is necessary at Y. All interior grid intersection points
at tg + At are computed in the manner described above. At
x = 0, the receding characteristic Eq. 5 is solved simultaneously
with the boundary condition equations. At x = L, the for-
ward characteristic Eq. 3 is solved with the downstream
boundary condition. When new values of flow and pressure
have been computed at each grid point at t = to + A t, time is
incremented to tQ + 2A t, and the solution is obtained at the
next higher set of grid points. The pipe end conditions to be
used in this problem include the injector needle at the down-
stream end of the fuel delivery line, the pump at the upstream
end of the fuel delivery line and also at the downstream end
of the fuel supply duct, and a constant pressure supply at the
upstream end of the fuel supply duct.
The equations to describe the dynamic action of the pump
and fuel injector must include:
1. A description of the compressibility of fluid in the
various volumes, of the following form:
K =
AP
AV/V
(7)
where:
K = Bulk modulus of fluid
V = Volume of enclosure
2. An unsteady continuity equation written for each
volume: the net mass inflow into an enclosure at an instant
must be equal to the time rate of increase of mass within the
enclosure.
3. An equation of motion to describe the forces and dy-
namic action of mechanical parts such as the delivery valve:
2F = mdV/dt
(8)
where:
£F = Sum of all pressure, frictional, and spring forces in
the direction of motion
m = Mass of object
V = Velocity
4. An equation to describe the flow rate through the various
ports and orifices, of the form:
(9)
where:
Cd = Discharge coefficient
A = Instantaneous area of opening
(Pi ~ P2) = Pressure drop across opening
Pump equations, for the elements shown in Fig. 3, are as
follows. Eq. 7 and the continuity equation for each of the
feed chamber, pump chamber, and delivery chamber are,
respectively:
dpf Kf
dt
Q2-Q4)
10
-------
Delivery Chamber
Transducer no 2
Pumping Chamber
Relief Valv«
L57
Feed Chamber
Fig. 3 - Sectional view of injection pump
dt
ir
Eq. 8 applied to the delivery valve becomes
01)
(12)
and
Nozzle Upper
Chamber
Transducer no. 4
zle Lower
Chamber
Injection Chamber
Transducer no. 5
Fig. 4 - Sectional view of injection nozzle
Q4 = Cd4A4 8
(18)
dt
= V
(13) The flow Q^s is in the supply duct, defined by Eq. 3, and
Qzd *s 'n tne Delivery P'Pe and 's defined by Eq. 5. The quan-
tities S, k, r, f, and W refer to valve displacement, spring force,
spring stiffness, friction coefficient, and valve weight, respec-
(14) tive'y-
Fuel injector equations. Fig. 4, take a similar form. The de-
livery and nozzle upper chambers, nozzle lower chamber, and
In these equations the subscripts f, p, d, and v refer to the feed injection chamber relations are:
chamber, pumping chamber, pump delivery chamber, and de-
livery valve, while flows with numerical subscripts refer to . ^
orifice flows, Fig. 3, and are defined by Eq. 9: —^ = —
QI =CdiA1V2g(prPp)/r
Q2 = Cd A2V2g(P - pf)/T
05)
(16)
dp, K,
07)
11
-------
FW. MfTIHINO STAN
Fig. 5 - Diesel fuel injection system test equipment
Eq. 8 applied to the needle becomes
dVm
IT = W^ (AuPu + AlPl + AiPi - kn - rnSn ~ fnVn)
and
i!s
dt
= V,
(22)
(23)
The subscripts u, 1, i, and n refer to the upper nozzle and de-
livery chambers, the lower nozzle chamber, the injector
chamber, and the injector needle. The flow Q^u refers to the
end of the delivery pipe and is given by Eq. 3. The orifice
flows, Fig. 4, a re:
Q5 =
Q6 =
Q7 =
r pc)/7
(24)
(25)
(26)
and leakage by the needle given by
Q8 ' C8 Pu
(27)
In Eq. 26, pc is the pressure in the combustion chamber, and
in Eq. 27, Cg is a viscous flow coefficient and function of the
area of flow and fluid properties.
METHOD OF SOLUTION - The procedure for numerical
solution of the partial differential Eqs. 1 and 2 by the method
of characteristics is well documented in the literature (15), and
results from such computations have been verified experi-
mentally. To obtain an accurate simulation of the complete
system, the pump and injector are treated as boundary con-
ditions for the pipeline and are solved at the identical time
step as the pipeline. Due to the extremely rapid response time
of the boundaries, particular care must be exercised to be as-
sured of a true solution to each set of ordinary differential
equations (Eqs. 10-14 and 19-23). Neither an iterative nor
Runge-Kutta approximation could be successfully applied at
time steps that were economically feasible for the method of
characteristics. A modified predictor-corrector method by
Hamming (16) is used in this model. This procedure offers the
advantage of being able to reduce the step size within a fixed
characteristics method time step in order to achieve a desired
accuracy level. Thus, during portions of the cycle, when the
transient contains high frequency components, a relatively
small step size is used, whereas a step size equal to the charac-
teristics time step is adequate over the major part of the cycle.
The predictor-corrector method makes use of the system re-
sponse at earlier times; additionally it requires information
from the pipeline during the step change. The latter informa-
tion is available from the characteristics described in Eqs. 3-6.
With the characteristics method solution in the pipelines and
the characteristics interfacing with the predictor-corrector
solution of the boundary equations, a solution for the total
system response is possible. The forcing function for the
entire system is the specified cam motion which drives the
pump plunger.
During the transient response, if the pressure tends to drop
below vapor pressure at any point in the system, the pressure
at the point is set equal to vapor pressure, and a vapor cavity
is assumed to form instantaneously. A local mass continuity
condition is established to simulate the growth and decay of
the vapor cavity. When the cavity collapses, the solution pro-
cedure returns to normal computations.
12
-------
UJ
QC
13
CO
CO
LU
cc
Q-
•WHBWV» d^^^m
JHH H^W WSfl M'
|135° I 18 O5
PUMP INLET CHAMBER
( Tr ansducer No 1 I
RESIDUAL LINE PRESSURE
(Strain Gage Transducer)
180°
PUMP DELIVERY CHAMBER
(Transducer Mo 2)
135°
,180°
PIPE END
(Transducer No 3)
INJECTION UPPER CHAMBER
(Transducer No 41
I 135° ,'165°
INJECTION CHAMBER
(Transducer No 5'
Fig. 6 - Oscilloscope reco/ds of transient phenomena in injection system
13
-------
Fig. 7 • Pump plunger motion and pump port areas; rack con-
trol setting = 0.509 in.
(40
145 150 155 160 165
PUMP CAM ANGLE. DEGREES
175
Jo
180
EXPERIMENTAL EQUIPMENT,
INSTRUMENTATION, AND RESULTS
The experimental apparatus and test results are described,
followed by a comparison of computed and experimental re-
sults. The purpose of the equipment is to determine quanti-
tatively the time variant response of the predominant variables
in the apparatus during a normal operating cycle. Practically,
it is impossible to measure all variables, nor is it necessary.
One of the primary objectives is to provide enough informa-
tion to confirm the analytical simulation of the operating sys-
tem. As already discussed, the analytical model must treat
each component part in great detail. A satisfactory confirma-
tion is acknowledged if the simulation provides agreement
with the experiment in a random sample of the dependent
variables. To accomplish this objective in a given test, the sys-
tem geometric configuration and fluid properties must be iden-
tified, and the pump speed and volumetric fuel supply setting
must be known. The system response is then identified by
transient pressure measurements at specific locations; a mea-
surement of fuel input at the pump and a measurement of fuel
injected at the nozzle. The latter measurements are averaged
over a number of cycles with the difference between the two
quantities representing the system leakage.
A general view of the diesel injection test bench is shown in
Fig. 5. This shows the injection pump mounted on a stand.
The pump is an American Bosch APEIB type and has a plunger
diameter of 10 mm. The pump is connected to the injector
by a high pressure pipe line 0.067 in. inside diameter, 0.25 in.
outside diameter, and 36.4 in. long. The injector is an Ameri-
can Bosch ADB-150S type, and has a 4-hole nozzle of 0.012
in. diameter each. The pump drive system includes a substan-
tial flywheel to provide uniform rotation and is driven by a
variable speed d-c motor. The speed is variable within the
range, 100 rpm-1000 rpm. The fuel rack is positioned by a
micrometer that has a travel between 0 and 1 in., correspond-
ing to maximum fuel and minimum fuel delivery, respectively,
with a minimum graduation of 0.001 in. Refs. 17 and 18 were
very helpful in the design and construction of the test equip-
ment.
The net fuel flow to the pump is measured using a metering
stand (left side of Fig. 5) which consists of a balance, a relay
system and an electric clock. The injected fuel is collected and
weighed. The fuel pressure is measured using AVL piezo
quartz pressure transducers. Fig. 1 shows the location of the
five pressure transducers. Further details of the transducer
location connections are shown in Figs. 3 and 4. An additional
strain gage-type static pressure transducer is positioned on the
pipeline directly above the pump. This unit was constructed
and added to the pipeline to measure the fuel line residual
pressure during tests. A Tektronics S02A type dual beam os-
cilloscope with a polaroid camera was used to record the pres-
sure traces.
The cam angles were measured every 3 deg, using a disc with
120 equally spaced holes mounted on the flywheel and an
electro-magnetic pick up, type 3010-AN, (made by Electro-
Mation Co.). The injector opening pressure for all tests was set
at 3000 Ib/in. , under quasi-static conditions. The fuel used
in these tests was Standard Oil Co. No. 2 diesel fuel.
A sample of the recorded pressure traces are displayed in
Fig. 6, representing a pump speed of 800 rpm, rack micro-
meter setting of 0.509 which corresponds to a fuel injection
rate of 0.19 Ib/min, and system base pressure (residual line
pressure) of 1400 psi. The upper line in each frame is the pres-
sure trace taken at each transducer location shown in Fig. 1.
The lower trace correlates the pressure trace with the instan-
taneous position in the pumping cycle. The plunger motion
and the area of the spill port are shown for this case in Fig. 7.
The experimental data obtained with this equipment are
very reproducible. That is, if identical test conditions are set
up on different days, identical results are produced. This in-
cludes not only base pressure and injected volume in each
cycle, but also individual pressure spikes in the recorded trace,
which may at first appear to be random. After-injection and
temporary conditions of vapor pressure are also very repro-
-------
20.08
K0.06
30.04
or o-
10000-
5000-
0-
10000-
5000-
.DELIVERY
VALVE
• •X ix x
VBLVEOPEN
XXX*
I
EXPERIMENTAL DATA
COMPUTER MODEL
RESULTS
•«X X X X X X
PUMPING CHAMBER
DELIVERY CHAMBER
NO. 2
PIPE END
NO. 3
NOZZLE UPPER
NO. 4
130
-10.016
QOI2
O008
0004
-0
-oooo
-5000
INJECTION CHAMBER
NO. 5
140 150 I
PUMP CAM ANGLE, DEGREES
10000
5000
-0
Fig. 8 - Comparison of injection system hydraulic characteristics - experimental and computer results, 800 rpm pump speed and O.S09 rack mi-
crometer (0.1910 Ib fuel injected/minute)
-------
z:0.08-
5o.o4-
6000
4000
2000h
Ql
eoooh
DELIVERY
VALVE
VALVE OPEN
x*xv
X
X
COMPUTER
NEEDLE
DKKXXXXXXXX XXXXXXKXXX x xxxxx
RESULTS -JQ000
Q(
0
PUMPING CHAMBER
DEUVERY CHAMBER
NO. 2
PIPE END
NO. 3
NOZZLE UPPER
NO 4
-6000
-4000
-2000
INJECTION CHAMBER
NO. 5
WO 150 160 170
PUMP CAM ANGLE, DEGREES
-6000
-4000
-2000
-0
160
Fig. 9 - Comparison of injection system hydraulic characteristics - experimental and computer results, 400 rpm and C.509 rack micrometer
(0.0968 lb fuel injected/minute)
16
-------
S' 0.08
K 0.06
5 0.04
0.02
0
600O
4OOO-
2000-
0-
6000-
6000-
DELIVERY
VAU/E .
VALVE OPEN*
TiTTx x x x «
a xx*
EXPERIMENTAL OATA-|o.oi6 ~
, V:"" COMPUTER MODEL
- NEEDLE
XXXXXXXX.XXXXXXXXXXXXKX
130
PUMPNG CHAMBER
DELIVERY CHAMBER
NO. 2
PIPE END
3
NOZZLE UPPER
NO. 4
140 150 160
PUMP CAM ANGLE, DEGREES
INJECTION CHAMBER
NO. 5
0.012 t
- 0-008 jj
-CX0043
"
-0
-6000
4000
200O
-6OOO
-4OOO
-2000
170
180
Fig. 10-Comparison of injection system hydraulic characteristics-experimental and computer results, 800 rpm and 0.675 micrometer (0.0818
Ib Cual Injected/minute)
17
-------
EXPERIMENTAL DATA
COWUTERMOOQ.
RESULTS
I4O ISO I6O 17O 180 I9O 200
PUMP CAM ANGLE, DEGREES
20 Z20
Fig. 11 • Comparison of pressures in pump delivery chamber, experi-
mental and computer results for a large cam angle
Table 1 - Comparison of Base Pressure* and Injected Fuel Quantity
Obtained from Experiment and Computer Model
Tigs. 7 and 8 Fig. 9 Figs. 10 and 11
Speed, rpm 800
Micrometer setting, in.
Range: 1 in. (no load) to 0.509
0 in. (full load)
Experiment, vol/cycle, in. 0.00796
liase pressure, psi 1400
Simulation, vol/cycle, in.3 0.00789
Uascpressure.psi 1600
400
0.509
0.00805
0.0084
2000
800
0.675
0.00340
1100
0.00357
1490
"The base pressure is the residual fuel-line pressure between
successive injections.
ducible if these are traits peculiar to the particular operating
condition.
COMPARISON-EXPERIMENTAL
AND ANALYTICAL DATA
Comparisons are presented for three different test condi-
tions. Fig. 8 displays computer results of pressure variations
at five locations, as well as pump delivery valve and nozzle
needle motion, all as a function of cam angle. Experimental
data are also shown for corresponding transducer locations.
These are the same data as displayed in Figs. 6 and 7 and rep-
resent the portion of the cycle when fuel injection occurs.
Figs. 9 and 10 provide comparisons of two different test con-
ditions: 400 rpm and rack micrometer of 0.509 in., and 800
rpm and rack micrometer of 0.675 in., respectively. A com-
parison of experimental and simulated pressure in the pump
delivery chamber for a larger pump cam angle is shown in
Fig. 1 1. These results are from the same test as those shown
in Fig. 10. Additionally, Table 1 provides a comparison of the
base piessure and fuel injected for each of the reported cases.
It is observed that the analytical model produces a reason-
ably accurate reproduction of the pressure-time response re-
corded from the experimental test setup. By reference to
Fig. 8, a cause and effect study of the response during the in-
jection period is possible. For example, the delivery chamber
pressure rises while the delivery valve opens, and drops with
the delivery valve when the plunger helix is uncovered. The
needle valve opens after the nozzle upper chamber pressure
exceeds the injector opening pressure of 3000 psi; the nozzle
upper chamber pressure drops temporarily and the injection
chamber pressure begins to rise at the instant of needle open-
ing; nozzle pressures and needle motion during after-injection
are easily correlated; etc. Such features as the time delay in
pressure wave travel from pump to injector needle, wave re-
flection phenomena, pressure peaking at the needle, delivery
valve over-shoot before stabilizing at a particular opening
during delivery, needle lift motion to the limit position and
holding during injection, after-injection 10-1 5 camshaft rota-
tion deg after the cessation of the primary injection, as well
as many other relevant phenomena are identifiable with care-
ful study of these figures.
DISCUSSION
The general agreement in pressure response between experi-
mental and simulation results generates some confidence in
the analytical model. The reproducibility of experimental
results produces a similar confidence in the reliability of the
measured quantities. Various features relative to both the
equipment and simulation are discussed in some detail.
MASS CONTINUITY • The experimental equipment pro-
vided a measurement of the fuel pumped and the fuel injected
over a number of cycles during the test. The leakage that
occurs at the needle in the injector and at the pump plunger is
available by subtraction. In most tests ihis leakage was less
than 3% of the volume injected. The computer model simi-
larly accumulated the volume pumped and injected during the
cycle and also the leakage volume. A good check on the re-
liability of a numerical simulation of an unsteady fluid flow
problem is the gross continuity balance. In acceptable simula-
tions of test runs with this model, the continuity was within
4.0%. The leakage volume per cycle is one of the important
variables to be accurately modeled, if a good pressure agree-
ment is to be expected. An additional check on the agreement
between experiment and model is provided in the mass conti-
nuity comparison, Table 1, where it is seen to be less than
4,5%.
BASE PRESSURE - The initial pressure in the delivery pipe-
line in the system is recognized as one of the important vari-
ables and, unfortunately, a difficult value to measure. The
establishment of an identifiable reference pressure is not easy
in the operating condition. If vapor pressure is reached and
can be identified it provides a suitable reference. In ihis appa-
ratus, the strain gage mounted on the delivery pipe was used
to establish this residual pressure. It is believed that the ac-
curacy of this reading is likely within 300 psi.
In the computer model an initial base pressure was used to
-------
start each tun. Initially this was always the experimental
value. Following the simulation, an evaluation of the elastic
energy stored in the system is possible by use of the pressures
along the pipeline, in the pump delivery chamber, and in the
injector. If ihe elastic energy stored after the cessation of fuel
supply and injection is different from the initial value, a new
value of base pressure is estimated and the run is repeated. No
more than one extra trial was needed to satisfy this condition,
and normally the correction was less than 300 psi. It is to be
noted that discrepancies in the initial and final base pressures
give rise to continuity errors in the simulation.
WAVE PROPAGATION VELOCITY - Th* effective bulk
modulus of elasticity of the fluid in the initial operating con-
dition was evalualed by use of a measurement of the wave
propagation velocity from the actual experimental recordings.
The variation of the modulus with pressure was consideied in
accordance with the literature (14). This refinement pro-
duced a wave speed variation between 4017 and 4391 fps in
the pipeline. Minor improvements in the timing of pressure
peaks were made with this modeling as compared with a con-
stant wave speed or bulk modulus.
DISTRIBUTED FRICTION • Viscous losses in the system
control the rate of attenuation of the residual waves in the
system following the primary injection, and exhibit only a
modest influence on the initial pressure and discharge waves.
The variation of the friction factor with Reynolds number in
the model is not considered too important, although too large
a pipeline resistance parameter drastically alters the response
pattern.
DISCHARGE COEFFICIENTS - Each of the valve ports,
orifices, and flow passageways requires a discharge coefficient
in the simulation that is a function of a number of parameters,
some of which are time dependent. The magnitude of (his
value at some of the locations is very critical in the production
of a reasonable simulation. Computational results seemed to
be most sensitive to the values used at the injector nozzle, the
needle passage to the injection chamber, and the pump de-
livery valve. These same values are also difficult to estimate on
the basis of information in the literature due to the unusual
geometric configuration. Steady-state measurements on this
system were not attempted. Best results in the simulation
have been obtained by use of values calculated, where possible,
from the measured transient pressure response. Values ascer-
tained in this manner have been reasonably constant in the
various experiments. Constant values have been used in all
past simulations.
VAPOR PRESSURE - When vapor pressure is reached in the
pipeline, a vapor pocket develops and subsequently collapses.
This phenomenon is discernible in transducer records by a flat
bottom pressure trace for a short period of time. The com-
puter model reproduces the same condition by not permitting
the pressure to drop below vapor pressure. A local cavity is
allowed to develop and its size is computed by a mass conti-
nuity balance. When the cavity goes to zero, the nornrJ comp-
utations resume with conditions of a homogeneous fuel at the
former cavity location.
SOLUTION OF DYNAMIC EQUATIONS - The handling of
the dynamic equations that describe the motion of the pump
delivery valve and the nozzle needle, and the fluid compressi-
bility in lumped volumes, is most critical to a satisfactory sys-
tem simulation. The response time of these features is ex-
tremely rapid. Hamming's modified predictor-corrector
method was found to be the most successful procedure in this
simulation.
CONCLUSIONS
The analytical model can be confidently used to study
various system configurations and proposed design changes.
The computer simulation can serve as an excellent design tool
with which to perform many additional parametric studies to
ascertain the influence of changes in various geometric and
mechanical variables. The analytical model can assist in under-
standing fuel injection systems, and its use can shorten the de-
velopment time required to improve fuel injection character-
istics which may lead to improved combustion and reduced
exhaust emissions. It will be apparent that the model de-
scribed can be most easily adapted to systems similar to the
one used in this work. We believe the modeling techniques
outlined can be applied effectively to other types of injection
systems.
ACKNOWLEDGMENT
This investigation was supported by United States Public
Health Service Research Grant No. AP0083S, from the
National Air Pollution Control Administration. The work was
also assisted by research funds provided by the Chevron Re-
search Co. The valuable contributions of graduate students
Bela Petry, Arthur Bauer, and S. Onyegegbu are also appreci-
ated.
NOMENCLATURE
= Pipe cross-sectional area, or area of opening at
orifice
= Wave propagation velocity
= Discharge coefficient
a
Cd
D
d
f
f
f
g
i
K
k
I
m
n
P
P
= Pipe diameter
= Subscript, refers to pump delivery chamber
= Friction coefficient at valve or needle
= Darcy-Weisbach friction factor
= Subscript, refers to pump feed chamber
= Acceleration of gravity
= Subscript, refers to injector chamber
= Bulk modulus of elasticity
= Spring force
= Subscript, refers to nozzle lower chamber
= Mass of valve or injector needle
= Subscript, refers to injector needle
= Pressure
= Subscript, refers to pump chamber
19
-------
Q = Volumetric flow rate
r = Spring stiffness
S = Valve or needle displacement
t = Time
u = Subscript, refers to upper nozzle and delivery
chamber
V = Velocity
V = Volume of fluid enclosures in pump and injector
v = Subscript, refers to delivery valve
W, X, Y, Z = Subscripts, refer to positions in the x-t plane
W = Weight of valve or needle
x = Distance
Zd = Subscript, refers to instantaneous flow in de-
livery pipe at pump delivery chamber
Zs = Subscript, refers to instantaneous flow in supply
pipe at pump feed chamber
Zu = Subscript, refers to instantaneous flow in de-
livery pipe at injector
7 = Specific weight of fluid
REFERENCES
1. S. J. Davis and E. Giffen, "Injection, Ignition; and Com-
bustion in High Speed Heavy Oil Engines." Proc. Instn. of
Automobile Engrs., (March 1931) p. 399.
2. Kalman J. De Juhasz, "Graphical Analysis of Transient
Phenomena in Linear Flow." Jrl. Franklin Inst., (April-June.
1937), pp. 463, 643, 655.
3. E. Giffen acid A. W. Rowe, "Pressure Calculations for
Oil Engine Fuel-Injection System." Proc. Instn. Mech. Engrs.,
London, Vol. 141 (1939) p. 519.
4. B. E. Knight, "Fuel Injection System Calculations."
Proc. Inst. Mech. Engrs. (A.D.), No. 1 (1960-1961).
5. P. I. Becchi, "The Analytical Investigation of Phenom-
ena Concerning the Fuel Injection in Fast Diesel Engines,
Carried out at Design Stage by Means of the Electronic Com-
puter." Tech. Bulletin, Fiat, Vol. XV, No. 2 (April 1962).
6. G. W. Brown and H. McCallion, "Simulation of an In-
jection System with Delivery Pipe Cavitation Using a Digital
Computer." Proc. Instn. Mech. Engrs., London, Vol. 182
(1967-1968) p. 206.
7. J. A. Stone, "Discharge Coefficients and Steady-State
Flow Forces for Hydraulic Poppet Valves." Jrl. of Basic Eng.,
Vol. 82 (I960) p. 144.
8. E. Van Walwijk, R. Van der Graaf, and J. K. M. Jansen,
"Berekening Van brandstofinspuitsystemen Voor dieselmoto-
ren." Werktuig-En Scheepsbouw, May 10, 1969, W 95.
9. Frank Kreith and Raymond Eisentadt, "Pressure Drop
and Flow Characteristics of Short Capillary Tubes at Low
Reynolds Number." ASME Trans. 56,1956.
10. A. Lichtarowicz, R. K. Duggins, and E. Markland, "Dis-
charge Coefficients for Incompressible Non-Cavitating Flow
Through Long Orifices." Jrl. Mech. Engr. Science, Vol. 7,
No. 2 (1965).
11. A. G. Gelalles, "Coefficients of Discharge of Fuel In-
jection Nozzles for Compression-Ignition Engines." National
Advisory Committee for Aeronautics Tech. Report No. 373
(1931).
12. P. Henrici, "Elements of Numerical Analysis." New
York: John Wiley & Sons, Inc., 1964.
13. J. B. Maxwell, "Data Hand Book on Hydrocarbons;
Application to Process Engineering." New York: D. Van
Nostrand Cq., Inc., 1950.
14. W. A. Wright, "Prediction of Bulk Moduli and PVT Data
for Petroleum Oils." ASLE47 AM 7B-l,May 1-4, 1967.
15. V. L. Streeter and E. B. Wylie, "Hydraulic Transients."
New York:'McGraw-Hill Book Co., 1967.
16. W. Ralston and H. S. Wilf, "Mathematical Methods for
Digital Computers." New York: Wiley & Sons, 1960, pp. 95-
109.
17. A. Bassi, "Experimental Investigations into Diesel In-
jection Systems." Sulzer Research No. 1963.
18. American Bosch Corp., "Fuel Injection and Controls
for Internal Combusion Engines." 1963 and New York:
Simmons Boardman Publ. Corp., 1962.
This paper is subject to revision. Statements and opinions
advanced in papers or discussion are the author's and are
his responsibility, not the Society's; however, the paper has
Society of Automotive Engineers, Inc.
been edited by SAE for uniform styling and format. Discussion will be printed
with the paper if it is published in SAE Transactions. For permission to publish
this paper in full or in part, contact Jhe SAE Publications Division and the
authors.
16 page booklet.
Printed in U.S.A.
20
-------
FLEET OWNER
MOTOR & EQUIPMENT MANUFACTURERS ASSN
SOCIETY OF AUTOMOTIVE ENGINEERS, INC.
Two Pennsylvania Plaza, Nevy York, N.Y. 10001
• ' ?^^Bkpr" ' '
Analysis and Control of
Transient Flow in the
Diesel Injection System
Part I -
The Analytical Control Method
Mohamed F. EI-Erian, E. Benjamin Wylie, and Jay A. Bolt
University of Michigan
SOCIETY OF AUTOMOTIVE ENGINEERS
Combined Commercial Vehicle Engineering & Operations
and Powerplant Meetings
Chicago, III.
June 18-22,1973
730661
21
-------
TABLE OF CONTENTS
Abstract
Comparisons of Experimental Data and Simula-
tion Program Results
Average Elastic Eneigy -The Control Parameter
Description of the Design Method on the Pipe
Characteristic Plane
Effect of the Injection System Average Elastic
Energy Function on the System Transient Pressures . .
Avcrigc lll^itii; Hr.cr^/ lUlvu.,.. Vi.. Acldi;^..<.! Cv,..lro!
Valve
Average Elastic Energy Release Via the Pump
Spill Port
Conclusions
Nomenclature
Acknowledgment
References
Appendix
Copyright ©Society of Automotiva Engineers, Inc. 1873
All rights reserved.
22
-------
730661
Analysis and Control of
Transient Flow in the
Diesel Injection System
Part I -
The Analytical Control Method
Mohamed F. EI-Erian, E. Benjamin Wylie, and Jay A. Bolt
University of Michigan
THE OPERATION of the injection system of a diesel engine
is of critical importance since it has a major influence on the
combustion process in the engine. Optimal operating con-
ditions are particularly important in view of current air pollu-
tion problems. Also, present uses of the diesel engine are
somewhat limited due to current operating combustion
characteristics.
One of the most persistent injection system problems is
related to wave phenomena in the high-pressure line between
the pump and nozzle assembly of the most popular type of
injection system. These pressure waves commonly result in a
secondary opening of the fuel injection nozzle following the
normal injection which is referred to as after-injection. This
results in reduced engine performance and increased emissions.
Bradbury (I )* showed that after-injection invariably
occurred at conditions of simultaneous high engine speed
and load. In addition, his studies on the relation between the
'Numbers in parentheses designate References at end of
paper.
ABSTRACT
The increasing requirements imposed on diesel engine
manufacturers have required the study of fuel injection
system faults and the development of means to eliminate
them. Until now, improved injection system characteristics
have been obtained by experimental trial-and-error proce-
dures. These procedures, however, have proved to be in-
convenient, tedious, and have had limited success in eliminat-
ing system faults such as after-injection. This is mainly because
the transient nature of the injection process requires a more
thorough study of the system time-varying parameters.
In this paper the residual transients which cause after-
injection are analytically investigated. The control of these
transients required specification of some system parameter.
The rapidly varying nature of the system pressures and flows
prevented the use of these variables as control parameters.
On the other hand, the average instantaneous elastic energy
stored in the system was found to be a well-behaved function
of time and was therefore selected as the control parameter.
A design procedure to control after-injection was formulated.
In this procedure, design changes required to achieve im-
proved system characteristics were obtained by specification
of the average elastic energy function. Two design changes
were considered. These included the design of an additional
control valve in the pump delivery chamber and the redesign
of the pump spill port. The design techniques used in the
control method are described in ihis paper, while the appli-
cation, results, and experimental verifications are given in a
companion paper.
-------
STRAIN
GAUGE
TRANSDUCER
DELIVERY CHAMBER
TRANSDUCER
WZZLE UPPER
CHAMBER
WOZZLE LOWER
CHAMBER
TRANSDUCER
INJECTION CHAMBER
l?ig. 1 - Schematic representation of diesel injection system and points of
pressure measurement
brake specific fuel consumption and break mean effective
pressure indicated ihe need for large injector holes to achieve
large output, and that small holes were more adequate for
part-loads. He concluded that small holes are preferred over
the whole range if it is possible to eliminate after-injection at
high outputs.
Similar information was obtained by personal communica-
tions with A. C. Rosselli (2): "We have consistently found that
high injection pressures with small holes tend to give superior
performance. The effect of very high injection pressures on
smoke tend to be especially significant." Furthermore, it is
shown in Ref. 3 that after-injection considerably increases
tinburned hydrocarbons emissions.
Mansfield (4) summarized the problem as, "It is difficult
if not impossible, for a pump to perform satisfactorily on a
heavy.duly engine under all conditions of speed and load, and
that the high rates of pressure change, which must occur in
all injection systems, lead to secondary injection."
Many experimental trial-and-error methods have been used
to control this phenomenon. These include studies of the
effect of the nozzle injection area and the delivery pipe
diameter or length among other parameters. Limited success
is expected from using these methods due to the fact that
these design changes are time-invariant and hence do not suit
the transient nature of the problem. More success may be
realized by design alterations of the time-varying parameters
of the system. Ail example of such a parameter is the
instantaneous area of the spill port during the spill period.
Recently Dolenc and Lees (5) and Lustgarten and Dolenc
(6) used a pump plunger with a stepped helix which provided
a throttling effect during the early part of the fuel spill period.
Their experimental results showed that the modified system
provided improved performance compared with the original
system. This was evidenced by less after-injection and re-
duced cavitationaJ damage.
Prior to 1960, theoretical studies of objectionable transients
in hydraulic systems were limited due to the requirement of
lengthy mathematical computations. By introducing a certain
number of simplifying assumptions, simple diesel injection
systems were analyzed and design variations tried until
acceptable performance v/as achieved. Since 1960 a number
of injection simulation computer programs have been
developed (7-9); however, none of the published literature in
this field indicates a use of the concept of design synthesis.
In related areas of fluid mechanics a procedure referred to
as "valve stroking" (10) has been developed and applied to
various water, oil, and natural-gas systems. In this procedure
the desired time-varying flow and pressure relations are
specified at a section in the system, and computations arc
performed to yield the required behavior at other positions
in the system. Boundary conditions may then be imposed to
accomplish the desired conditions. This synthesis procedure
is a more direct approach to design changes in transient fluid
flow problems and leads to sophisticated control techniques
not easily attainable by standard simulation methods.
Although direct application of the valve-stroking techniques
to the diese! injection system proved unsuccessful, the cal-
culation concepts were applicable. Failure of the direct
-------
Test
No
I
2
3
4
Table 1 • Resume of Testing Conditions Used to Comput with the Simulation Program
and Used to Study the Af ter-lnjection Phenomenon
Speed.
rpm
800
365
•105
700
Rack Micrometer
Setting, in*
0509
0.608
0.350
0426
'Rack micrometer setting: in = I -
Total Fuel
Injected.
Ib/min
0.191
0.054
0.148
0.216
lest load
maximum load
Base
Pressure.
PSI
1400
1300
2000
2500
O.OOOP8
0.000127
0.000126
0.000128
After-Injection.
Ib/min
0.0510
0.0055
00343
0.0796
% After-Injection
of Total Injection
26.70
10.19
23.18
36.85
* "C( = cd • Ah where A^ is the area of nozzle injection holes and cd is a coefficient of discharge.
£ O.O8
£0.06
30.O4
d 0.04
IOOOO
VWVEOPEN
5000-
IOOOO-
5000-
EXPERIMENTAL DATA
COMPUTER MODEL
RESULTS
• PUMP
.DELIVERY
VALVE
-0.0*
INJECTOR.'
.NEEOLEX
PUMPING CHAMBER
DELIVERY CHAMBER
NO. 2
NOZZLE UPPER
NO. 4
INJECTION CHAMBER
NO. 5
140 155 H
PUMP CAM ANGLE,
-IOOOO
50OO
0
-IOOOO
SOOO
180
DEGREES
Fig. 1 - Comparison of injection system hydraulic characteristics-experimental and computer results. 800 rpm
pump speed and 0.509 tack micrometer (0.1910 Ib fuel injected/min)
-------
0.016 u.
EXPERIMENTAL DATA
INJECTION CHAMBER
NO. 5
140 145 150 155 160
PUMP CAM ANGLE. DEGREES
165
Fig. 3 • Comparison of injection system hydraulic characteristics-experimental and computer results, 365 rpm
pump speed and 0.608 rack micrometer (0.054 Ib fuel injected/min)
Table 2 - Comparison Between Theoretical and Experimental Results for Data
Presented in Table 1 and Figs. 2-5
Experimental Data
Simulation Results
Test
No.
1
2
3
4
Speed,
rpm
800
365
405
700
Rack Micrometer
Setting, in
0.509
0.608
0.350
0.426
Injected Fuel,
in /cycle
0.00796
0.00493
0.01218
0.01029
Base Pressure,
psi
1400
1300
2000
2500
Injected Fuel,
in /cycle
0.00789
0.00511
0.01193
0.01011
Base Pressure,
psi
1600
1400
1988
.2430
26
-------
?008
-" 0.06
^004
IJOO2
3 0
8000
4000
0
in
a-8000
LU
o:
o
$4000
Id
-------
-0.08
I 0.06
-1 0.0.4
=> O-
8000
4000
0
PUMP
DELIVERY'
VALVE •
— EXPERIMENTAL DATA
~.m COMPUTER MODEL
.RESULTS
X
' INJECTOR •/"
. .' NEEDLE "T
PUMPING CHAMBER
DELIVERY CHAMBER
NO. 2
NOZZLE UPPER -
NO. 4
INJECTION CHAMBER
NO. 5
135 140 !45 150 155 160 165 170
PUMP CAM ANGLE, DEGREES
0.016 t
0.012 J
0.008 y
0.004 g
0 "
12000
8000
4000
0
12000
8000
4000
175 ISO
Fig. S - Comparison of injection system hydraulic characteristics-experimental and computer results, 700 rpm
pump speed and 0.426 rack micrometer (0.216 Ib fuel injected/min)
range of speed and load conditions. Four of these test runs,
with different amounts of after-injection, are summarized in
Table 1. Each test is identified by pump cam speed, a fuel
rack setting, the total amount of fuel injected, the pipe base
pressure which is the residual pressure remaining in the
pipeline between cycles, and the coefficient Ct which is the
product of the total nozzle injection hole area and the
coefficient of discharge. In addition, the amount of after-
injection and its percentage to the tota) injection are given
in Table 1.
Tests 2, 3, and 4 were run at a lower injection hole area
than test 1. This was necessary to produce after-injection at
various speed and load conditions so that this phenomenon
could be simulated and then theoretically investigated. Tests
2, 3, and 4 were chosen to represent low-, medium-, and
high-speed and load conditions, respectively. It is seen from
the table that a simultaneous increase of speed and load re-
sults in an increase of after-injection.
Comparisons between experimental data and results from
the simulation program for the four test conditions listed in
Table 1 are presented in Figs. 2-5. Comparisons of transient
pressure traces are possible at four locations: the delivery
chamber, the pipe end, the nozzle upper chamber, and the
injection chamber. Also, theoretical results for the pressure
variation at the pumping chamber and motion of the pump
delivery valve and nozzle needle are included in the same set
of figures. Ml the injection system variables are presented as
functions of the pump cam angle for the portion of the cycle
in which fuel injection occurs. The computer model results,
together with the experimental data for the system residual
. pressure and the injected fuel for each of the four cases con-
sidered, are listed in Table 2. The favorable comparisons
between the results from the analytical model and the ex-
perimental data generate confidence in the underlying as-
sumptions and techniques used in developing the model.
A cause-and-effect study of the injection system response
28
-------
is possible by examining any of the presented cases (Figs. 2-5).
For example, (he pressure in the pumping chamber begins
to rise slowly because the ports are still partly uncovered.
When the ports are covered, the pressure rises rapidly. The
pressure in the pump deliver)' chamber rises when the
deliver)' valve opens and drops with the delivery valve clos-
ing. The nor/.le needle opens after the pressure in the nozzle
upper chamber exceeds the injector-opening pressure of
3000 psi. This causes the nozzle upper chamber pressure to
drop temporarily and the injection chamber pressure to
begin to rise at the instant of needle opening. The needle
motion and nozzle pressure are easily correlated for after-
injection. The wave reflection phenomenon and the time
delay in pressure wave travel are made clear by comparing
pressures at different locations.
AVERAGE ELASTIC ENERGY-THE
CONTROL PARAMETER
In this study, the term average elastic energy is used to
define the average pressure stored in the injection system at
an instant of time. It should be noted that this pressure has
the units of energy per unit displaced volume at the
boundaries. Two examples are given to illustrate this defini-
tion. First, a moving frictionless piston in a liquid-filled
cylinder exerts work on the liquid. This energy is stored in
the elastic deformation of the liquid and could be used to
drive the piston back to its initial position. The second
example is drawn from unsteady liquid flow in a frictionless
pipe. In this example, elastic energy is stored in a liquid
column due to the compression of the column. This energy
is released as kinetic energy by the expansion of the column.
Trie dotted system in Fig. 6 is used to calculate the injec-
tion system average elastic energy. This system includes the
fluid in the pump delivery chamber, in the injector delivery,
upper and lower chambers, and in the delivery pipeline. The
terms P and L-J-QJ are defined in Fig. 7. In this figure,
4*UMP anc^ HNJ are ecluiva'ent 'engths to match the
corresponding volumes in (lie pump and injector, respectively.
The average elastic energy at art instant is calculated in the
manner described in Figs. 6 and 7.
During each cycle the pump adds energy in the form of
flow work which is temporarily stored in (he system as elastic
energy. Part of this energy is released at the injection
nozzle during the injection period. If excessive uncontrolled
elastic energy remains in the system following the closure of
the pump delivery valve, it will be released at the nozzle
resulting in after-injection.
The average elastic energy together with pressure in the
pump delivery chamber and nozzle upper chamber for test
1 (Table 1) are displayed in Fig. 8. All the traces begin at
!he point when the spill port begins to open. In region !,
the average elastic energy drops because of spilling at the
pump spill port plus the energy release due to injection at
the nozzle. Region II begins when the delivery valve closes.
The energy drop in this region is due to injection at the
-SPILL POST
PLUNGER ^DELIVERY VALVE
AVERAGE ELASTIC ENERGY
L INJECTOR
WEEOLE
LTOT
Fig. 6 - Schematic representation of system used to calculate she average
elaslic energy
nozzle only. Region II ends when the nozzle needle begins
to close while region 111 coincides with the period of needle'
closure. For this region slight increase in elastic energy
might occur due to the effect of work done by the needle
compressing the fuel during its downward motion. When the
needle finally closes, the nozzle upper chamber pressure is
lower than the needle-opening pressure. However, the inter-
mediate elastic energy in region IV is high, thus causing the
pressure at the nozzle upper chamber to increase as the
pressure wave reflects in the closed system. When this pres-
sure reaches the needle-opening pressure, the needle opens
and the system elastic energy drops during region V. Later
on, the upper chamber pressure drops below the needle-
closing pressure and the needle closes. At this point, the
system elastic energy finally reaches the residual energy. The
elastic energy stays constant during region VI until the next
cycle begins.
The average elastic energy trace in Fig. 8 gives a better
understanding of the after-injection phenomenon. A study
of a number of similar cases led to the conclusion that the
manner in which the energy is released during the spill period
(region 1) largely regulates the behavior of the residual
transients in the injection system. For example, a rapid
energy drop in region 1 causes a rapid closing of the delivery
valve which means a higher intermediate energy and there-
fore an increased after-injection. On the other hand, a very
slow drop during the spill period leads to prolonged injection
periods and low-pressure injection. Tims one possible ap-
proach to the reduction of undesirable transients in the
system is to focus attention on the controlled release of tlie
average elastic energy.
DESCRIPTION OF THE DESIGN METHOD
ON THE PIPE CHARACTERISTIC PLANE
The investigation of a specific case of after-injection begins
with recording experimental data of the actual test condition.
Then the simulation program is used to simulate this condi-
tion. Comparisons between theoretical results and experi-
mental data of four cases of after-injection were presented
in Figs. 2-5. The design program presented in trus section
serves the purpose of finding means (o eliminate aiter-injec-
29
-------
AVEWfiE ELASTIC ENERGY*
LTOT~\
PUMP DELIVER* CHAMBER-
INJECTOR
DELIVERY
CHAMBER
' i-TOT"
-au
Fig. 7 - Sketch showing the method foi calculating the injection system average elastic energy
tion. This is achieved by utilizing a controlled release of the
stored elastic energy in the system. The resulting alterations
in pressure and flow patterns are calculated and translated
into feasible design changes using the design program.
Fig. 9 represents the pipe characteristic plane (x-t plane).
An instant of time is represented by a horizontal line while a
vertical line represents a specific location along the pipe. The
line AJ represents the location of the pump delivery chamber,
and the line Bl represents the location of the injector delivery
chamber. Line ED represents the instant of time at which the
negative wave, resulting from the uncovering of the pump
spill port, reaches the injector. Line HG represents the in-
stant of nozzle needle closing. Three important events
characterize the performance of the diesel injection system
arid are given in Fig. 9: First, the plunger helix begins to un-
cover the spill port at point C; then the pump delivery valve
is totally dosed at F; finally, the nozzle needle is totally
closed at G. In this figure, the distance AC is the wave travel
time and BD is twice the wave travel time.
The design method is illustrated in Fig. 10 which is similar
to Fig. 9 but includes more details of the solution procedure.
In this figure, the distance between the vertical lines repre-
sents a pipe section of length Ax1, while the distance between
the horizontal lines is the characteristic method time step
At. The data needed by the design program include a
complete description of the injection system geometric
configuration and properties. Simulation program results of
pressure and flow along the line BD are specified for the
design program. Also a control period beginning at point C
together with a controlled average elastic energy function
are specified. The choice of the length of the contrpl period
and the energy function during this period depend on the
injection system properties and the choice of the controlling
device. Further discussion of this subject will be presented in
the next section.
Several regions are given in Fig. 10 to simplify the presenta-
tion of the solution method. The solution procedure in
region I begins with known pressure and flow conditions at all
points along the vertical line BD. Conditions at points lying
on another vertical line adjacent to BD are calculated by the
use of the pipe characteristic equations (Eqs. A-l 1 and A-l 3
in the Appendix) in a manner illi/strated by the triangle
ZjYjWj. In this triangle, pressure and flow conditions are
known at Yj and W^ but are not known at Zj. It should be
noted that Zj Wj is the C
characteristic line and YjZj is the
C" characteristic line. When the conditions at all points
lying on a vertical line are calculated, the .calculation pro-
ceeds to points on the next adjacent vertical line. Calculations
in region I end when the pressure and flow at point C afe
known.
In region H, calculations begin at t = L/a + At. At this
instant, pressure and flow conditions are known for ail' pipe.
locations except the boundary point- W^. Conditi6ns at this
boundary point are calculated by making use of the assumed
instantaneous average elastic energy together with an equation
from the pipe C" characteristic line. At t = L/a + 2At,
pressures and flows are known at all points except at Zj anc!
Z-j. At this point of time, conditions at the interior point
-------
_ 8000
w
Q_
£ 6000
tc
LU
5 4000
o
2000
8000
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
Needle Needle
Reopen Reclosed
PUMP DELIVERY CHAMBER
NOZZLE UPPER CHAMBER
Needle Opening
_ Pressure
8000
6000 ^
uT
cc
40003
2000
150 155 160 165 (70
PUMP CAM ANGLE, DEGREES
Fig. 8 - Transient pressures and average elastic energy versus pump cam angle for
test 1 (Table 1)
CHAMBER
Pig. 9 - Injection system performance on the pipe x-t plane
(Z2) are calculated by utilizing the known pressures and flows
at W2 and Y2 together with the pipe characteristic equations
along the C* line W2Z2 and the C" line Y2Z2- Conditions at
Zj are calculated in the same manner described for W2. This
procedure continues by advancing to the next step, calculat-
ing the interior points, and then calculating the boundary
point along the line CE. Calculations in this region end by
calculating pressures and flows at all points along the horizon-
tal line ED.
Calculations in region 111 proceed by calculating all
pressures and flows at time t prior to calculating these con-
ditions at time t + At. This is done by first calculating the
interior points in the same manner previously described for
point Z2 in region II. Then conditions at the injector
boundary point along the line DG are calculated by solving
the injector equations (Eqs. A-6-A-10) together with the
pipe characteristic equation along the C+ line illustrated by
-------
c
1
0.1
•ol
CONTROL PERIOD.
^SPECIFIED ELASTIC,
y
4—
F
S E
cc t-
LU
JJ
r
W
!ix
z.
^^
x.
X
z.
<9y£^
•JS^
•**^
^
^^^^
v»
^\^
z.
tf<
'8
^^
RE
3ior
RE(
RF<
^^
V8
RE
>ION
?|ON
«•
^
3IOf
•V
j EZ
m
Tt
j i
w«
^ — '
Z,
t
'^^
z.
^
^4
X
X
\^
w,
^
^
Y,
^
^
t
O
f
C
»-
L
2
«•
C
EQUATIONS c
i
3i
a
O
lit
- — SPECIFI
*
2L
B
x-0 x-L
Fig. 10 - Design method on the pipe x-t plane
the line W^Z^. Finally the pump boundary point along EH
is calculated. During the control period, this calculation is
performed in the same manner described for Wj. After the
end of the control period, the compressibility equation
dP/dt = -
-------
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
PUMP DELIVERY CHAMBER
NOZZLE UPPER CHAMBER
ISO
155 160 165 170
PUMP CAM ANGLE, DEGREES
i-'ig. 12 - Injection system variables versus pump cam angle, liffect of timing the
average elastic energy drop. Control valve approach, test 1 tTable 1)
the desired optimum function is defined as the one that will
result in damping the residual transient pressures in a way
that eliminates after-injection over a desired range of operat-
ing conditions without seriously altering the injection pres-
sures during the main injection period and without introduc-
ing any undesirable effects (such as cavitation). This definition
implies Ihe following:
1. The original average elastic energy1 function during
region I (Fig. 8) should not be seriously altered in order to
retain the shape of the injection pressures during the main
injection.
2. The injection system stored elastic energy must reach
the system original residual energy (Fig. 8) by the end of the
control period shown in Fig. 10.
Different objective functions could be defined and utilized
equally well to assist in the solution of other types of injec-
tion design problems
In order to physically achieve a desired specified elastic
energy function, two design changes were considered. The
first is the addition of a control valve in the pump delivery
chamber, as shown in Fig. 11. This valve must open and
release any excess elastic energy' at the correct time and
with the correct rates in order to damp the transient pressures
and eliminate after-injection. The second alternative is to
redesign the pump spill por! and pump delivery valve to
insure that the excess elastic energy is released at the pump
during the spill period.
AVERAGE ELASTIC ENERGY RELEASE
VIA ADDITIONAL CONTROL VALVE
"Die control valve concept introduced in this paper is
similar to the duplex or two-way delivery valves which have
been experimentally investigated. This paper contributes a
calculation procedure which can be used by the design
33
-------
_8000
IE
£6000
oc
LJ
54000-
2000
12,000
10,000
_ 8000
Q.
w- 6000
cc
84000
IT
O.
2000
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
PUMP DELIVERY CHAMBER
NOZZLE UPPER CHAMBER
-8000
6000 a?
4000|
LLl
2000 ?
150 155 160 165 170
PUMP CAM ANGLE, DEGREES
175
Fig. 13 • Injection system variables versus pump cam angle. Effect of the shape
of the average elastic energy drop. Control valve approach, test 4 (Table 1)
engineer to determine when and how this valve should be
operated.
An arbitrary release of fuel through a valve during the
desired control period is likely to be unsuccessful as far as
satisfying the objectives. The following few alternatives
point out some of the complications. Figs. 12 and 13
illustrate the interaction of the time-variant average elastic
energy and pressure transients in the system. All traces in
these figures begin at the beginning of the control period.
The solid traces represent the system performance without
alterations. The appearance of high residual energy in the
solid traces and the existence of negative pressures in some
of the traces are due to the fact that after-injection and
cavitation were not accounted for in the design program.
Two alterations are given in Fig. 12 to illustrate the
dependence of pressure on the time of initiating the elastic
energy drop. The results given in this figure represent test
1 (Table 1 and Fig. 2). The circles represent an elastic energy
drop starting at the point of delivery valve closing while the
triangles represent an energy drop starting at the point of
needle final closing. From comparisons of resulting pressures
at the nozzle upper chamber, the following is clear: The
first alteration (given by circles) results in added cavitation.
Also, after-injection is liable to occur since the residual
pressures at the injector are not damped. The second altera-
tion (given by triangles) results in damping the residual
pressures and therefore after-injection will not occur. Also,
no cavitation is detected in the second example. Thus it
can be concluded that an elastic energy drop beginning at
the point of final needle closing is desirable. Several other
trials have confirmed this conclusion.
Several shapes of elastic energy drop were studied and three
of them are given in Fig. 13. Results in this figure are for
test 4 (Table 1 and Fig. 5). In this figure, all alterations
-------
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
PUMP DELIVERY CHAMBER
NOZZLE UPPER CHAMBER
135 160 165 " 170"
PUMP CAM ANGLE. DEGREES
Fig. 14 • Injection system variables versus pump cam angjc. Effect of tlic rate or'
the average elastic energy drop. Redesigned spill port approach, lest 4 (Table 1)
begin at the point of final needle closing. The three elastic
energy alterations are represented by dots, triangles, and
circles. By comparing pressures at the nozzle upper
chamber, it is clear that the average elastic energy function
represented by the circles results in damping the residual
pressures and thus eliminates after-injection. The other
two alterations reduce the magnitude of the resulting pressures
but do not completely eliminate the possibility of after-
injection.
AVERAGE ELASTIC ENERGY RELEASE
VIA THE PUMP SPILL PORT
The second approach deals with redesigning the pump spill
port. The object of the new design is to release the high
elastic energy in the system by controlling the flow of fuel
from the spill port. To allow this, the pump delivery valve
must be open during a relatively long control period. This
can be done by using slower rates of energy release and by
employing delivery valves with high unloading volumes.
Several choices of average elastic energy functions are given
in Figs. 14 and !5. Results in both figures are for test 4
(Table 1 and Fig. 5). Traces on these figures are used to
illustrate the effect of the average elastic energy on the
transient pressures in the system. Three of these functions
are given in Fig. 14 and are used to illustrate the effect of
changing the rate of elastic energy drop near the end of the
control period on the resulting transient pressures, it is
seen from this figure that the solid trace represents a desirable
solution; the cases represented by circles and triangles are
undesirable. The triangles represent a very slow drop of the
elastic energy function. "Die very slow drop results in high
residual pressures at the nozzle upper chamber which will
prolong the main injection period. The circles represent a
relatively fast elastic energy drop. This change results in
high nozzle upper chamber pressures winch lead to after-
injection. Tne solid trace of the elastic energy function
results in damping the residual pressures and eliminating the
possibility of after-injection.
Fig. 1 5 is intended to illustrate the effect of the manner
of variation of average elastic energy on the resulting transient
pressure. Three functions are given to demonstrate this
effect. Al! three functions in this figure have the same be-
ginning and end points for the average elastic energy function.
The solid average elastic energy ttace represents a desirable
solution since it results in damping the nozzle upper chamber
-------
•. V* AVERAGE ELASTIC ENERGY
IN THE SYSTEM
PUMP DELIVERY CHAMBER
NOZZLE UPPER CHAMBER
155 160 i65~J70
PUMP CAM ANGLE. DEGREES
175
Fig. 15 - Injection system variables versus pump cam angle. Effect of the shape of
the average elastic energy drop. Redesigned spill port approach, test 4 (Table 1)
pressures, therefore eliminating after-injection. The other
two traces (triangles and circles) have less damping effect
on the transient pressures than the solid line trace. Also,
the trace given by triangles will result in prolonged injection
period due to the relatively nigh pressures near the end of
injection.
Usually no more than four trials were needed to achieve a
desirable average elastic energy trace for both designs. It
should be noted that the procedures described here for the
formulation and application of the design program are
general and could be used with any injection system. How-
ever, the results presented here apply only for the specific
injection system described in this study.
CONCLUSIONS
The computer simulation used in this study accurately
imitates the physical system behavior and therefore may be
used confidently to study the diesel injection system. A
critical evaluation of the various phenomena which influence
performance is possible by parametric studies.
Injection system average elastic energy is shown to be a
suitable and convenient parameter to be utilized for studies
of design improvements. Unlike the very erratic time-variant
pressure and flow variations in the-system, the average elastic
energy is a smooth function of time. Design variations to
eliminate after-injection are realizable, and a theoretical
procedure is presented to accomplish this objective. The
implications of two possible design alterations are discussed.
Futher amplification of the use of control devices and both
theoretical and physical demonstrations of their ability to
accomplish the defined objectives appear in the companion
paper.
The design procedure described herein may be applied to
other types of injection systems and to the solution of other
injection problems. Additionally, the methods have applica-
tion to other interactive fluid and mechanical systems which
are characterized by undesirable high-frequency oscillations.
NOMENCLATURE
a
C,,
pipe cross-sectional area or area of opening at
orifice
wave propagation velocity
discharge coefficient
-------
Ct = product of the nozzle injection holes area and
the coefficient of discharge
D = pipe diameter
d = as a subscript, refers to pump delivery chamber
f = friction coefficient at valve or needle
f = Darcy-Weisbach friction factor
f = as a subscript, refers to pump feed chamber
g = acceleration of gravity
h = as a subscript, refers to nozzle injection holes
i = as a subscript, refers to injection chamber
INJ = as a subscript, refers to equivalent pipe length
of injector volume
K = bulk modulus of elasticity
k = spring force
1 = as a subscript, refers to nozzle lower chamber
n = as a subscript, refers to injector needle
p = pressure
p = as a subscript, refers to pumping chamber
PUMP = as a subscript, refers to equivalent pipe length
of the pump delivery chamber volume
Q = volumetric flow rate
r = spring stiffness
S = valve or needle displacement
t = time
TOT = as a subscript, refers to total pipe length includ-
ing equivalent lengths of volume at pump and
injector
u = as a subscript, refers to upper nozzle and
delivery chamber
V = velocity
V = volume of fluid enclosures in pump and in-
V jector
v = as a subscript, refers to delivery valve
W, X, Y, Z = as subscripts, refer to positions in the x-t plane
w = weight of valve or needle
x = distance
Zd = as a subscript, is the instantaneous flow in the
delivery pipe at the pump delivery chamber
Zs = as a subscript, is the instantaneous flow in the
supply pipe at the pump feed chamber
Zu = as a subscript, is the instantaneous flow in the
delivery pipe at the injector
y = specific weight of fluid
ACKNOWLEDGMENT
The financial support of the Environmental Protection
Agency through research grant No. R800424 is gratefully
acknowledged. This work was also assisted by research
funds provided by the Chevron Research Co. The valuable
assistance of S. Cmyegegbu, PhD student at the University of
Michigan, is also appreciated.
REFERENCES
1. C. H. Bradbury, "Stationary Compression Engines."
London: E. & F. N. Spon. Ltd., 1950.
2. A. C. Rosselli, Manager of Controls and Fuel Systems,
Cummins Engine Co., personal communication, July 13,
1972.
3. Jay A. Bolt and M. F. El-Erian, "Diesel Ignition,
Combustion and Emissions." Progress Report No. 19 sub-
mitted to the U.S. Army-Tank Automotive Command,
Warren, Michigan, September 1972.
4. W. P. Mansfield, "A New Servo-Operated Fuel In-
jection System for Diesel Engines." SAE Transactions, Vol.
74 (1966), paper 650432.
5. A. Dolenc and R. Lees, "Some Aspects of the Develop-
ment of the OVA24 Traction Engine." Proc. Inst. Mech.
Engrs., Vol. 183, Part 3b, 1968-1969, p. 24.
6. G. Lustgarten and A. Dolenc, "Development of In-
jection System for Medium-Speed Diesel Engine with
Quiescent-Type Combustion Chamber." Diesel Combustion
Symposium, Inst. Mech. Engrs., London, April 1970.
7. G. A. Becchi, "Analytical Simulation of Fuel Injection
in Diesel Engines." SAE Transactions, Vol. 80(1971),
paper 710568.
8. B. E. Knight (M.A.), "Fuel Injection System Calcula-
tions." Proc. Inst. Mech. Engrs. (A.D.), No. 1, 1960-1961.
9. E. B. Wylie, J. A. Bolt, and M. F. El-Erian, "Diesel
Fuel Injection System Simulation and Experimental Corre-
lation." SAE Transactions, Vol. 80 (1971), paper 710569.
10. V. L. Streeter and E. B. Wylie, "Hydraulic Transients."
New York: McGraw-Hill Book Co.. 1967.
11. M. F. El-Erian, "Simulation and Control of Transient
Flow in the Diesel Injection System." PhD thesis, University
of Michigan, April 1972.
-------
APPENDIX
THE INJECTION SYSTEM EQUATIONS
The equations describing the injection system are divided
into three categories: pump equations, injector equations,
ami pipe equations. These equations are summarized below.
Further information concerning these equations and their
solutions can be found in Refs. 9 and 11.
PUMP EQUATIONS
The pump equations written for the feed chamber, pump-
ing chamber, and delivery chambers (shown in Fig. A-l) are
given below, respectively.
dpf
(A-l
(A-2)
dt
Each of the above equations is a combination of continuity
and compressibility equations written for a particular volume.
In addition, the delivery valve equations are given by:
dV,,
= JE-,
d, w -(Av(Pp-Pd>- V'vV fvVv)
dSv
'dt
(A-4)
(A-5)
INJECTOR EQUATIONS
These equations take a similar form and are written for the
injection delivery and nozzle upper chamber, the nozzle power
chamber, and the injection chamber (Fig. A-2), respectively:
^Pu
dt
r(Qzu-
dt
dp,
dt
(Q6-0-7-Aivn) (A'8) VZA'
Additionally, the needle motion is described by the following
two equations:
dt
w ^AuPu * AlP] * Vi - *n - rnSn ' fnVn>
Fig. A-l - Sectional view of injection pump
Fig. A-2 - Sectional view of injection nozzle
(A-9)
A C B L Length X
Fig. A-3 - Characteristics in x-t plane
-------
dt
= V,
(A-10)
2 DA
= 0
PIPE CHARACTERISTIC EQUATIONS
The equations are written in finite difference form for a pipe
section of length X and a time period t (Fig, A-3) as follows:
-------
, „ ,:
Analysis and Control of
Transient Flow in the
Diesel Injection System
Part II - Design Results of
Controlled After-Injection
Mohamed F. El-Erian, E. Benjamin Wylie, and Jay A. Bolt
University of Michigan
SOC I ETY OF AUTOMOTIVE EN G I N EERS
Combined Commercial Vehicle Engineering ft Operations
and Power plant Meetings
Chicago, III.
June 18-22,1973
.
730662
-------
TABLE OF CONTENTS
Abstract
Calculation of Design Changes at the Pump
The Additional Control Valve Design •
The Redesigned Pump Spill Port
Design Results and the Computer Simulation Confirmations of the
Additional Control Valve Design . . .
Design Results and the Computer Simulation Confirmations of the
Redesigned !'t:^p Spill Port
Experimental Verification of the Redesigned Pump Spill Port Approach,
Experimental Modifications
Comparisons Between Experimental Results of the Original and
Modified Systems
Conclusions
Nomenclature
Acknowledgment
References
Copyright ©Society of Automotive Engineers, Inc. 1973
All rights reserved.
-------
730662
Analysis and Control of
Transient Flow in the
Diesel Injection System
Part II - Design Results of
Controlled After-Injection
Mohamed F. EI-Erian, E. Benjamin Wylie, and Jay A. Bolt
University of Michigan
MANY DIESEL FUEL injection problems are related to the
wave propagation phenomena in the pipeline connecting the
pump and the injector. Most of these problems such as after-
injection, cavitation, and needle impact may be controlled by
controlling fuel release after the end of the main injection.
However, due to the complexity of the fuel injection system,
past control methods have largely depended upon cut-and-try
experimental procedures and upon the use of many simplify-
ing assumptions in analytical studies.
Several experimental procedures have been used to attack
the problem of after-injection. Guglieimotti (1)* used differ-
ent types of vented pump delivery valves to reduce the injec-
tion system residual pressure. However, the reduced residual
pressures resulted in cycle-to-cycle irregularities of injection
characteristics at low-speed and load conditions. Lustgarten
and Dolenc (2) were able to assume quasi-static conditions for
*Numbers in parentheses designate References at end of
paper.
After-injection is the introduction of additional fuel to the
combustion chamber after the end of the main injection. It
is a persistent diesel fuel injection problem which usually
results in reduced engine power and economy and increased
emissions. After-injection is caused by uncontrolled pressure
transients at the injector after the opening of the pump spill
port. These pressure transients are related to the wave propa-
gation phenomena in the high-pressure pipeline connecting the
pump and injector. Use of experimental trial-and-error
methods in attempts to control this phenomenon has met
with limited success.
The analytical control method described in another paper
is used to determine design means by which after-injection
-ABSTRACT
may be controlled. Further investigation and evaluation of
two design changes which release the injection system excess
elastic energy in a controlled manner are considered herein.
One design change is the addition of a control valve in the
pump delivery chamber. The other is the modification of
the pump spill port. In both cases, pressures and flows are
not altered during the main injection period. The ability of
both design changes to control after-injection is confirmed
by use of a simulation program. Experimental data from a
system with the'pump spill port modified in accordance with
theoretical design calculations provide satisfactory confirma-
tion of the analyses.
-------
theii injection system which was characterized by short fuel
lines iind relatively small dead volumes. Their model neglected
the effect of fluid elasticity and wave propagation in the high-
pressure pipeline. Under these assumptions they were able to
calculate a pump spill port area which resulted in reducing
after-injection while maintaining desirable pipe residua!
pressures.
In pan 1 of this paper (3), the injection system stored elastic
energy was investigated as a useful design control parameter.
A control technique was described in which improved injection
characteristics were obtained by specifying the stored elastic
energy in a controlled manner. In this second part, results of
studies of iwo improved designs are presented. The first is
the addition of a control valve in the pump delivery chamber.
This valve opens a! the correct instant and releases the excess
elastic energy in a controlled manner. In the second design,
after-injection is eliminated by redesigning the pump spill port
and pump delivery valve to control the rate of fuel spill. In
both designs the original system pressure and flow conditions
are not altered during the main injection period.
A computer simulation program is used to confirm the
control method results. Also, calculated design modifications
of the spill port and pump delivery vajve are used to modify
an actual injection system. Experimental comparisons
between the original and modified injection system pressure
characteristics are given to verify the validity of the design
technique predictions.
CALCULATION OF DESIGN
CHANGES AT THE PUMP
The design procedure described in part 1 of this paper (3)
utilized an assumed average elastic energy profile and known
initial conditions to calculate pressure and flow conditions in
the whole x-t plane given in Fig. 10 (part 1). To make use of
these calculated conditions, they must be translated into
feasible design changes in the injection system.
Two design changes at the pump, aimed at the elimination
of after-injection, were suggested. The first design is a con-
trolled relief valve in the pump delivery chamber which opens
to release the excessive elastic energy at the correct instant.
The second is a redesign of the pump spill port area to
release the elastic energy in a controlled manner. In both
designs, the calculated pressure and flow data at the pump-
pipe boundary (p^, Q^) during the control period can be
used as added boundary conditions to the pump equations
(Eqs. A-l-A-5, part I). The solution of these equations
provides the information to accomplish the desired design
alterations. The equation and solution methods used in
both designs are considered below.
THE ADDITIONAL CONTROL VALVE DESIGN - The
control valve used in the first design is illustrated in Fig. 11
(part 1). In order to maintain the injection pressures during
the main injection period, it is assumed that this valve will
open to release the excess elastic energy after the pump
delivery valve closure. Therefore, the equation representing
the pump delivery chamber (Eq. A-3, part I) is modified to:
dt
where:
p = pressure
Q = volumetric flow rale
K = fluid bulk modulus of elasticity
V = volume
t = independent variable of time
d = as subscript, refers to the pump delivery chamber
c.v. = the additional control valve
Zd = boundary at the pump delivery chamber and the pipe
intersection
The flow through the control valve (Q v ) is given by:
Qc.v. - Cdc.v.Ac.v.
(2)
where:
Cd = coefficient of discharge
A = area
g = acceleration of gravity
7 = weight of fluid per unit volume
The calculated pressures and flows (pj and Q^A) at the pump-
pipe boundary during the control period are used in Eqs. 1 and
2 together with a reasonable value of the valve coefficient of
discharge of 0.7 to give the needed time-variable area of flow
through the valve. Results of the computed area variations are
discussed later.
THE REDESIGNED PUMP SPILL PORT - The second
design determines a pump spill port area from the known
pump-pipe boundary conditions (PJ, Qvj)- Tm's >s done by
solving the pump delivery chamber equation (Eq. A-3, part I)
together with the delivery valve motion equations (Eqs. A-4
and A-5, part I). Each of these equaiions may be written in
the following form:
dP
dt
(3)
where P is the dependent variable on the left side of the above
equations and 0 (t) is a function of time representing the right-
hand side of these equations. Eq. 3 may be written in the
following finite difference form:
where the subscripts 1 and 2 refer to the beginning and end of
the time period At. The left-hand side is the rate of change of
the dependent variable with respect to time, and the time
function on the right-hand side is calculated at the end of the
-------
155 I6O 165
PUMP CAM ANGLE. DECREES
175
Fig. 1 - Design program results of injection system variables and
simulation confirmation of these results. The additional control
valve for test 1 (Table 1, part 1)
period. It was found that the use of average values for the
right-hand side during the period resulted in divergence of the
solution due to the highly varying nature of the delivery valve
velocity during the closing period. Extreme variations in the
delivery valve velocity during the closing period result from
the use of a light valve spring and the fact that the acting
pressure force (P(j-pp) is small and is highly varying. Con-
verging solutions were obtained by using the finite difference
form given by Eq. 4.
In the finite difference forms of Eqs. A-3-A-5 (part I),
p^ and Q^jj are known from previous calculations, and p_, sv,
and Vv are the unknowns to be determined. These equations
are solved by using an iterative procedure and a convergence
technique described in Ref. 4.
The calculated pressure in the pump chamber (p ) together
with the calculated valve motion (sy, Vy) are used in the finite
difference form (Eq. 4) of the pumping chamber equation
(Eq. A-2, part I) to determine the spill port area. For this
equation, the feed chamber pressure (pj-) is assumed at a value
of 150 psi on the basis of previous simulation results. It
should be noted that the spill port area is highly insensitive to
the value of pj- because p^-is much less than the pumping
chamber pressure (p ) during the major part of the spill.
8000-
£6000
5 4000
••-DESIGN PROGRAM
SIMULATION PROGRAM
CONTROL VALVE
00004-
00002 S
-0
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
8000
5i6000
a.
u»
§4000
ffzooo
PUMP DELIVERY CHAMBER
NOZZLE UPPER CHAMBER
0-
152
156 160
PUMP CAM ANGLE. DEGREES
8000
6OOO I
I
I
4OOO i
I
2000 '
164
Fig. 2 - Design program results of injection system variables and
simulation confirmation of these results. The additional control
valve for test 3 (Table 1, part 1)
DESIGN RESULTS AND THE COMPUTER SIMULATION
CONFIRMATIONS OF THE ADDITIONAL CONTROL
VALVE DESIGN
A study of the effect of timing and shape of the average
elastic energy function on the system residual pressures was
presented in part I. In this study, the average elastic energy
curve given by circles in Fig. 13 (part I) represented a desirable
solution for the after-injection problem. A control valve
appended to the pump delivery chamber to release the system
elastic energy in the controlled manner is one way to achieve
the desired result. The area of flow through the control valve
is calculated by the design program in the manner described by
Eq. 2.
Design program results and simulation program confirmations
of these results for tests 1,3, and 4 (Table 1, part I) are given
in Figs. 1, 2, and 3, respectively. In these figures the circles
represent results obtained by the design program, while
simulation program confirmations of these results are illus-
trated by solid traces. The traces include the system average
elastic energy, pressures in the pump delivery chamber and
injector upper chamber, and the area of flow through the
control valve. It should be noted that a simple valve motion
-------
poooe
AVERAGE ELASTIC E
IN THE SYSTEM
fc&OOO
150 155 160 165 1TO
PUMP CAM ANGLE, DEGREES
!7ig. 3 • Design program results of injection system variables and
simulation confirmation of these results. The additional control
valve for test 4 (Table 1, part 1)
was used in the simulation program. As a result, the valve
area is described by a linear opening of the valve, a constant
maximum opening area, and a linear closing. This area was
obtained by averaging the design program results for the
control valve areas, as shown in Figs. 1-3.
In all the cases presented, the control valve starts to open
when the needle finally closes. The maximum opening area
differs from one case lo the other, the combined higher load
and speed conditions require larger areas of flow through the
control valve than the relatively lower speed and load condi-
tions. The valve-opening periods for tests 3 and 4 (Figs. 2 and
3) are both 1.2 ins. However, these periods are different in
terms of cam angle degrees due to the speed difference
between the tests. The valve-opening period for test 1 (Fig. 1)
is 2 ins. This is greater than for tests 3 and 4 primarily
because of the relatively low base pressure associated with this
lest*.
The average elastic energy traces in Figs. 1-3 result in the
damping of the system pressure oscillations and therefore the
elimination of after-injection. It should be emphasized that
§2000
H-" 008
—
^oo6
^ 004
£002
a
o
Q
6OOO
40OO-
; 2000
! 0
I
-200O
SYSTEM WITHOUT CHAMGES
SYSTEM WITH CHANGES
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
PUMP SPiLL PORT .
- OOO4
PUMP DELIVERY VALVE
PUMP DELIVERY CHAMBER .
NOZZLE UPPER CHAMBER
0012 S .
0003 o°
LJ
CE J
400O
200O;
This test uses a different nozzle area than tests 3 and 4.
115 ISO 155 ISO
PUMP CAM ANGLE. DEGREES
F:ig. 4 - Comparisons of original injection system performance and
design program results of the modified system. The redesigned spill
port for test 2 (Table 1. part 1)
the injection characteristics during the main injection period
are not altered since the average elastic energy function during
this period and the system base pressure are not changed.
Simulation program results confirm the design program
predictions. Small differences between design and simulation
results can be attributed to use of the simpler valve motion in
the simulation program.
DESIGN RESULTS AND THE COMPUTER SIMULATION
CONFIRMATIONS OF THE REDESIGNED PUMP SPILL
PORT
A study of the effect of the average elastic energy function
tor use with the redesigned pump spill port was presented in
part I. In this study, the solid line average elastic energy trace
in Figs. 14 and 15 represented a desirable solution to the
after-injection problem. This solution is achieved by redesign-
ing the pump spill port in order to release the system elastic
energy in the desired controlled manner. Design results of
the required spill port area together with the simulation pro-
gram confirmations of these results are now presented for tests
2-4 (Table I, part I).
The original system performance was illustrated in part I in
Figs. 3-5. A comparison between this performance and the
H6
-------
ff 8000
— SYSTEM WITHOUT CHANGES
SYSTEM WITH CHANGES
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
SPILL PORT
BEGINS TO OPEN
PUMP SPILL PORT -
PUMP DELIVERY VALVE
PUMP DELIVERY CHAMBER .
NOZZLE UPPER CHAMBER
165 160 165
PUMP CAM ANGLES, DEGREES
170
Fig. 5 - Comparisons of original injection system performance and
design program results of the modified system. The redesigned spill
port for test 3 (Table 1, pan I)
modified system as calculated by ihe design procedure is
given in Figs. 4-6 for tests 2-4, respectively. In these figures
the solid traces represent the original system performance
while the circles represent the modified one. Comparisons are
given for the system average elastic energy, the pump spill
port areas, the pump delivery valve motion, and for pressures
in the pump delivery and nozzle upper chambers. As men-
tioned previously, the design procedure utilizes a controlled
function for the average elastic energy to calculate the per-
formance of the modified system. It may be observed in the
figures that the design changes result in damping the pressure
oscillations thereby eliminating after-injection. Most of the
elastic energy release is accomplished at the pump by keeping
the pump delivery valve open for longer periods compared
with its original performance. This is achieved by use of the
relatively smaller areas of the modified spill port and by use of
a delivery valve that permits a higher unloading volume.
In all the cases, a delivery valve diameter of 0.374 in was
used instead of the original diameter of 0.236 in. It should be
noted that the resulting system base pressure for the modified
system is the same as the original one, and that the resulting
pressures at the injector for the modified system differ only
slightly from the original system.
It may be noted in Figs. 4-6 that the resulting area of the
pump spill port starts with a linear opening and then reaches a
— SYSTEM WITHOUT CHANGES
VN- SYSTEM WITH CHANCES
AVERAGE ELASTIC ENERGY
IN THE SYSTEM
155 160 165 i TO ITS
PUMP CAM ANGLE. DEGREES
ISO
Kig. 6 • Comparisons of original injection system performance and
design program results of the modified system. The redesigned spill
port for test 4 (Table 1, part 1)
constant vaiue. In Fig. 6, the spill port area begins to increase
again near the end of delivery valve closing. The constant
opening of the spill port is nearly the same in ail cases
(0.0011 in ). Therefore, it is possible to describe a common
spill port atea as a function of pump plunger lift for all the
reported cases. Such a common description is not possible in
terms of time or pump cam angle because of the difference in
speed for each test.
A common spill port area was used in the simulation
program in order to confirm the design procedure results. The
modified spill port areas are drawn as a function of pump cam
angle and are compared with the original areas in Figs. 7-9.
As mentioned previously, these modified areas would appear
identical if they were drawn as functions of pump plunger
lift. In the same set of figures, comparisons between the
injection pressure characteristics for the modified and original
systems indicate that slight differences occur for tests 2 and 3
while a relatively greater difference is observed with test 4. It
should be noted that test 4 has high-speed and load conditions
and high system base pressure (2400 psi). This relatively high
base pressure is higher than the needle-closing pressure (2200
psi). Therefore, it can result in prolonging the injection
period, as seen from Fig. 9. Better results for the system
injection pressures were obtained by using a different area
for the spill port, as seen in Fig. 10. The choice of this new
area was directed toward lowering the system base pressure.
This was achieved by using higher uncovering rates for the
-------
ry
£ 0 004
O y,
y, 0002
^ u.
30 0
6000
irt
O.
u' *»000
i/)
£ 2000
rr
0.
• SYSTEM WITHOUT CHANGES
- ooooo SYSTEM WITH CHANGES IN
THE SPILL PORT AREA
/
/ PUMP SPILL PORT
- ./
s o
X „ 0 0 °
^^ O
^°°°°°°
1 r^
(f O:
f*J \
J \ INJECTION CHAMBER
"" a ^Z°
j \l £\
1 ! 1. 1 i !
135 140 145 150 155 160 165
PUMP CAM ANGLE. DEGREES
l-'ig. 7 - Comparison ol injection chamber pres-
sures of original and modified systems. The
redesigned spil! port for test 2 (Table I, part I)
~ 0010
< z
a ~0008
< i-
g 00006
tt Q-
tf -J 0 004
O =
2 Si 0002
- 8000L-
a
SYSTEM WITHOUT CHANGES
oooooSYSTEM WITH CHANGES IN
THE SPILL PORT AREA
PUMP SPILL PORT
INJECTION CHAMBER
140
145 150 155 160 165
PUMP CAM ANGLE. DEGREES
lri&. 8 • Companion of injection chambei ps«-
^ sures of original and modified systems. The
redesigned spill port for test 3 (Table 1, part 1)
0.012
0.008
85;
'I 12.000
LJ
8000
4000
SYSTEM WITHOUT CHANGES
ooooooSYSTEM WITH CHANGES IN
THE SPILL PORT AREA
PUMP SPILL PORT
INJECTION CHAMBER
140
Fig. 9 • Comparison of injection chamber pres-
145 150 155 160 165 '~ 170 175 ISO sures of original and modified systems. The
PUMP CAM ANGLE, DEGREES redesigned spill pon for test 4 (Table 1, part 1)
-------
SYSTEM WITHOUT CHANGES
2 ~ O.OI2(- •••••• SYSTEM WITH CHANGES IN
THE SPILL PORT AREA
i- O.OO8 -
j
j
£ O OO4
* o
13.000
£ aooo
3
ft 400O
K
O.
PUMP SPILL PORT
INJECTION CHAMBER
135 140 145 150 155 160 165 J70 (75 J80
PUMP CAM ANGLE, DEGREES
Fig. 10 - Comparison of injection chamber pressures of original and modified systems. The
redesigned spill port for test 4 (Table 1, part 13
fc
on*
-J.OO4
a:
tn
U- /wa
ID AREA 0
UNCOVERI
, 1
x^
>"^ III 1 1 L. 1 1
MODIFIED
ORIGINAL
.02 .04 06 09 .10 .12 .14 .16 .IB
PUMP PUUNGCR MOTION AFTER-SPILL PORT OPENS, IN.
Fig. 11 - Comparison between the theoretically predicted spill area and
the experimentally used one
spill port near the end of injection while keeping these areas
the same as those in Fig. 9 for the early part of the spill. The
system base pressure for the results given in Fig. 10 was
found to be 1600 psi.
The modified spill port area associated with test 4 and given
in Fig. 10 was used in the simulation program to investigate
its effect on tests 2 and 3. This area resulted in lowering the
base pressures for these tests and also yielded a slight improve-
ment in the injection pressure characteristics. However,
extreme lowering of these base pressures at a combined low-
load and speed condition (test 2) results in reaching vapor
pressures. It should be noted that mechanical damage due to
vapor bubble collapse is less harmful at combined low-speed
and load conditions than at the combined high-speed and load
conditions (2). Such minor conditions of temporary existence
of vapor pressure may be tolerated if they only occur at low-
load and speed conditions.
Fig. 12 - Comparison between original and modified spill ports
EXPERIMENTAL VERIFICATION OF THE
REDESIGNED PUMP SPILL PORT APPROACH
The experimental system used in this study is a commer-
cially available American Bosch injection system which
consists of an APEIP type of pump and an ADB type of injec-
tion nozzle. This system is described in detail in Refs. 4 and
-------
MODIFIED
ORIGINAL
Fig. 13 - Comparison between original and modified pump delivery
valves
tu
I"
V
ft
UJ
o
Maximum
Valve
Original
Syittm
Modified -
Sytttm
.02
.Of, .00 .10 .12 .14
DELIVERY VALVE MOTION, IN.
.16
16
Fig. 14 - Fuel flow area past the delivery valve. Comparison of original
and modified designs
5. [t should be noted that this injection system is used by
International Harvester Co. on the Army Tank Automotive
Command research engine (the TACOM engine). This engine
is used at the University of Michigan for combustion and
emission research. Emission and performance studies on this
engine (6) showed the adverse effect of after-injection and the
need for eliminating it.
An experimental study was therefore carried out on the
injection system to verify the validity of the theoretical
results obtained from the application of the analytical control
technique. Further combustion and emission studies of
improved injection system characteristics are planned.
In the experimental study, after-injection was increased in
the experimental system by using a restricted nozzle hole area.
The original nozzle had four holes, 0.0128 in diameter each.
The injection area was then reduced by passing, a 0.010 in
diameter steel wire through two of the injection holes. Under
these conditions the product of the unrestricted injection area
and the coefficient of discharge (Cf = Cd • A^) was found to
be about 0.0002 in . Some variations in C{ of up to 10%
were noticed and related to inaccuracies in the method used to
reduce the injection area. The use of the above-reduced area
resulted in the presence of after-injection over a large portion
of the system-operating conditions. The control technique
was used with the above conditions to predict the required
design changes (spill port area and delivery valve diameter).
EXPERIMENTAL MODIFICATIONS - The predicted spill
pott area is shown in Fig. 11 as a function of the pump
plunger motion. This area consists of a linear opening,
followed by a constant area of 0.00151 in , and then another
linear opening to reach a maximum opening of 0.00303 in .
This was achieved by using a set of three circular holes. The
modified area is also given in Fig. 11. In addition, a compari-
son between the original and modified spill port designs is
given in Fig. 12. In the modified system, the lower two holes
are each 0.31 in diameter and are located at the same angle as
the plunger helix. These holes are used to achieve the first
area opening described in Fig. 11. After the lower two holes
are completely uncovered, the spill area remains constant at
^
0.00151 in until the plunger helix starts to uncover the upper
port. The upper port is 0.44 in diameter and is used to
achieve the second area opening given in Fig. 11. After all
three ports are uncovered, the spil! area remains constant at
0.00303 in2. It is clear from Fig. 11 that the actual design
modifications are very close to the theoretically desired
conditions.
The pump delivery valve was also redesigned in order to
insure the existence of an open valve during the control
period. It was found that the fuel flow areas past the delivery
valve could be kept the same as the original valve. However, a
larger valve diameter was needed to delay the valve closing
until the end of the control period. A comparison between the
modified and original delivery valves is given in Fig. 13. The
modified valve had a diameter of 0.374 in while the original
valve was 0.236 in in diameter. However, the fuel flow area
past the delivery valve was kept approximately the same as
the original system (Fig. 14) despite the use of a larger diameter
valve. This was achieved by using a taper of 30 deg after the
end of the valve collar, as shown in Fig. 13. Several other
modifications of the pump body, delivery valve seat, valve
spring, and pump delivery chamber were required to accom-
modate the new bigger valve. It should be noted that these
changes were designed in a way to maintain the same original
delivery chamber volume.
COMPARISONS BETWEEN EXPERIMENTAL RESULTS
OF THE ORIGINAL AND MODIFIED SYSTEMS • Experi-
mental data of the modified system were collected over a
-------
ORIGINAL SYSTEM
MODIFIED SYSTEM
f80°
PUMP DELIVERY CHAMBER
(Transducer No.2)
ISO"
1180° 1135
NOZZLE UPPER CHAMBER
(Tronsducer No.4)
I8O°
1135" '165
Pump Conishaft Angle Pump Camshaft Angle
INJECTION CHAMBCR
(Transducer No.5 >
1165°
Fig. IS - Comparison of injection system hydraulic characteristics, original and
modified systems, 600 rpm pump speed and 60% of maximum fuel rack
setting
wide range of speeds and fuel deliveries. These data were
compared with the data previously collected with the original
system. Samples of these comparisons are given in Figs. 15-17
for three cases of low, medium, and high after-injection of the
original system. In each of the figures, comparisons are given
between measured injection pressures at three different
locations*: the pump delivery chamber, the nozzle upper
chamber, and the nozzle injection chamber. All pressures in
the above figures are given as a function of the pump cam
angle.
*A schematic diagram of the injection system is given in
Fig. 1, part I.
It is clear from the comparisons that the modified system
transient pressures are controlled and damped after the end
of the main injection while the original system pressures are
relatively uncontrolled. After-injection was completely
eliminated from the modified system performance in both
Figs. 15 and 16, whiie a slight after-injection was present in
Fig. 17.
Two important injection system characteristics may be
noted from Figs. 15-17:
1. After-injection in the original system was caused by
uncontrolled pressure transients at the pump delivery cham-
ber. The pressure at the pump delivery chamber starts to
drop sharply when the spiil port opens. This pressure drop
ends when the pump delivery valve is closed. At this point.
-------
ORIGINAL SYSTEM
MODIFIED SYSTEM
(80°
PUMP DELIVERY CHAMBER
ETronsductr No. 2)
I8O" 1135
NOZZLE UPPER CHAMBER
(Transducer No.4)
ISO'
Pump Comihof t Angle
180"
INJECTON CHAMBER
(Tran»ductr No.5)
Pump Camshaft Arxjte
180°
Fig. 16 - Comparison of injection system hydraulic characteristics, original and
modified systems, SOO rpm pump speed and 60% of maximum fuel tack
setting
the wave reflections result in a sharp increase of the pump
delivery chamber pressure. This sharp increase is reflected in
the pressures at the nozzle upper chamber and results in
after-injection at the injection chamber.
2. The modified system pressures at the pump delivery
chamber are decreasing at a slower rate than the original
system; however, by the end of the spill period, the pump
delivery chamber pressures are completely controlled. The
slow drop at the pump delivery chamber is not reflected as a
slow drop in either the nozzle upper or the injection chamber
pressures.
It should be noted that it is widely accepted that a sharp
pressure drop at the injector can only be achieved by a sharp
pressure drop at the pump. Contrary to this belief, the
results presented in Figs. 15-17 illustrate that a sharp pressure
drop at the injector can be achieved while using a relatively
slower rate of pressure drop at the pump.
The design objectives defined in part I of this paper were:
1. To control the transient pressures after the end of the
main injection by damping them in a way to eliminate after-
injection,
2. To maintain the pressure and flow characteristics during
the main injection period,
Table 1 facilitates an assessment of the degree of success in
achieving the above objectives. This is done by a comparison
between important injection characteristics of both the
original and modified injection systems at the three conditions
presented in Figs. 15-17. In these comparisons, the maximum
difference in base pressures was less than 18%, the maximum
cycle pressure differed by less than 10%, the main injection
-------
ORIGINAL SYSTEM
MODIFIED SYSTEM
[80° H35
PUMP DELIVERY CHAMBER
(Transducer No 2)
160°
16O°
NOZZLE UPPER CHAMBER
(Transducer No 4)
ISO*
(TraniducerNo.5)
Fig. n - Comparison of injection system hydraulic characteristics, original and
modified systems, 900 rpm pump speed and 70% of maximum fuel rack
setting
Table 1 - Comparison Between
Injection System
Test Identification
Pump speed, rpm
Fuel rack setting as % of maximum
rack travel
Base pressure, psia
Maximum injection pressures, psia
Main injection period, deg
Fuel delivery during main injection,
in /cycle
Fuel delivery during after-injection.
in /cycle
Injection Characteristics of Both the Original and Modified injection Systems
at the Test Conditions Given in Figs. 1 5-1 7
Maximum
Original System
I
600
60
2200
8200
14.94
0-00868
0.00282
II
800
60
2000
9100
17.02
0.00800
0.00358
III
900
70
2200
9200
21.8
0.00947
0.0049 1
Modified System
I
600
60
1800
7400
15,45
0.00913
0
11
800
60
2200
8700
18.60
0.0089S
0
111
900
70
2400
8600
21.2
0.01027
0.00061
Difference
is less than
-
-
18%
10%
10%
12%
-
-------
period differences were less than 10%, and ihc fuel deliver)'
volume during main injection was within 12%. The above
differences are due to many factors which include variations in
(he no/./.le effective area and slight differences introduced by
ihe design modifications, among others.
CONCLUSIONS
After-injection results from uncontrolled pressure transients
after the end of the main injection period. Control of this
undesirable phenomenon is possible by controlling the rates
of pressure release at the pump. The injection system average
elastic energy was shown to be a suitable and convenient
control parameter which may be used to control undesirable
injection "pressure characteristics.
The control technique presented in this study allows a
theoretical systematic design improvement of diesel injection
characteristics. This was demonstrated by two improved
designs in which the injection system pressures after the end
of the main injection were controlled in order to eliminate
after-injection. Such knowledge may be helpful in improving
future injection systems and may lead to improved combustion
and reduced exhaust emissions.
A redesigned pump spill port or a control valve appended to
the pump delivery chamber may be used to eliminate after-
injection. Either approach, if carefully designed, may be
introduced without serious alterations in the injection system
performance during the main period of injection. It was
shown in this study that sharp pressure drops at the injector
can be obtained while using slower rates of pressure drop at
the pump.
Attempts to ev;duate combustion characteristics of the
newly developed system are being studied at the University of
Michigan. Reduced emissions and increased efficiencies from
using the modified injection system are anticipated benefits.
A magnitude of injection pressures unattainable in the old
system is also expected. Preliminary results indicate that after-
injection from this system will be less sensitive to reductions
in the injection holes area.
NOMENCLATURE
A = area of flow through the additional control valve
Cd = coefficient of discharge
C. = time-averuged flow coefficient equal to the product of
the nozzle holes area and its average coefficient of dis-
charge
c.v. = as a subscript, refers to the additional control valve
d = as a subscript, refers to the pump delivery chamber
f = as a subscript, refers to the pump feed chamber
g = acceleration of gravity
h = as a subscript, refers to the nozzle injection holes
K = bulk modulus of elasticity
p = pressure
p = as a subscript, refers to the pumping chamber
P = used as a dummy variable in Eqs. 3 and 4
Q - volumetric flow rate
s = deliver/ valve lift
t = time
v = as a subscript, refers to the pump delivery valve
V = velocity of pump delivery valve
V = volume
Zd = as a subscript, refers to the instantaneous flow in the
delivery pipeline at the pump delivery chamber
7 = specific weight of fluid
At = time step
0 = a time function representing a time rate of change at
the right-hand side of the pump equations
ACKNOWLEDGMENT
The financial support of the Environmental Protection
Agency through research grant No. R800424 is gratefully
acknowledged. This work was also assisted by research funds
provided by the Chevron Research Co. The valuable assistance
of S. Onyegegbu, PhD student at the University of Michigan,
is also appreciated.
REFERENCES
1. A. Guglielmotti, "Influence of the Construction Charac-
teristics of the Injection Pumps on Some Phenomena of
Instability and Irregularity of Diesel Engine Operation." Tech.
Bulletin, Vol. IX, No. 3 (July-September 1956), Fiat Co.,
Torino, Italy.
2. G. Lustgarten and A. Dolenc, "Development of Injection
System for Medium-Speed Diesel Engine with Quiescent-
Type Combustion Chamber." Diesel Combustion Symposium,
Inst. Mech. Engrs., London, April 1970.
3. Mohamed F. El-Erian, E. Benjamin \Vylie, and Jay A.
Bolt, "Analysis and Control of Transient Flow in the Diesel
Injection System, Part I-The Analytical Control Method."
Paper 730661 presented at SAE Combined Commercial
Vehicle Engineering and Operations and Powerplant Meetings,
Chicago, June 1973.
4. M. F. El-Erian, "Simulation and Control of Transient
Flow in the Diesel Injection System." PhD thesis, The
University of Michigan, April 1972.
5. E. B. Wylie, J. A. Bolt, and M. F. El-Erian, "Diesel Fuel
Injection System Simulation and Experimental Correlation."
SAE Transactions, Vol. 80 (1971), paper 710569.
6. Jay A. Bolt and M. F. El-Erian, "Diesel Ignition, Com-
bustion and Emissions." Progress report No. 19 submitted to
the U.S. Army Tank Automotive Command, Warren, Michi-
gan, September 1972.
"I his paper is suhject lo (
Su!enn'nt\ and opuuoras
advanced in papers or. discussion :,if the ^uiliot^ und .n
his tc-.pon«.\Uiiuy. \\\>\ the Society's. tu,v.evci. the pj|>cr h;ii.
been edited hy SAt for uniform siymu*, jiiU lormat. Dis-
fUSSMn will be pruned v-trh the p-p-'i i! il is published
Socieiy of Automotive Engineers. Inc.
in SAL TMn*-K[:oin. Tor jx'i minion 10 publish tins papei in full 01 m part,
coiiucl Ihc SAT Puhhoiioni Divismn
Poisons w]<.|;iii£ i>» submit pajirc. in he (.oMSiJetcd for pfrsrm.ition m put-l".a
UOM through SAt shuuKS sent! ihc hi.irumupl or 3 .100 u-md aliMiaci of a p'u
posed manuscript in Scuefary. L^iu'cung Activities tiojiJ. SAl:
16
c booklet
I'imleJ in U.S.A.
-------
TECHNICAL REPORT DATA
(Please read iMtmctlotts on the rev^ne before completing)
1. REPORT NO.
EPA-460/3-74-001
a.
A. TITLE AND SUBTITLE
Diesel Fuel Injection System Simulation and Experi-
mental Correlation
7. AUTHOR
-------
ENVIRONMENTAL PROTECTION AGENCY
Technical Publications Branch
Office of Administration
Research Triangle Park. N.C. 27711
POSTAGE AND FEES PAID
ENVIRONMENTAL PROTECTION AGENCY
EPA - 335
OFFICIAL BUSINESS
AN EQUAL OPPORTUNITY EMPLOYER
Return this sheet if you do NOT wish to receive this material fn,
or if change of address is needed j. (Indicate change, including
ZIP code.)
------- |