-
Eindhoven, University of technologyFaculty of Mechanical
EngineeringSection Procestechnical constructions
Measurement of particlesize distributions in Diesel
emission gasses
K.M.J Verschuur 414614
CompanionDr. ir. H.P. Van Kemenade
Eindhoven, 27th of March 2000
-
Measurement of particle size distributions in Diesel emission
gasses
2
Contents
1. Introduction 4
2. Particle size distributions in Diesel engines 5
2.1 Diesel engines 52.2 options for separation techniques 6
3. the measurement technique 8
3.1 Operating principles 83.2 Particle size distributions 103.3
Experimental parameters 11
4. Spheriglass and smoke experiments 12
4.1 The Diesel emission test rig 124.2 Pressure calculations
134.3 The Spheriglass experiments 14
4.3.1 the test rig 154.3.2 results 15
4.4 Smoke experiment 164.5 Explanation of possible problems
16
5. Diesel emission experiment 18
6. Conclusion and recommendations 19
Literature 20
Appendix 1. Geometric loss factors 21Appendix 2. Numerical
results of the measurements 23Appendix 3. Fraunhofer theory 27
-
Measurement of particle size distributions in Diesel emission
gasses
3
List of symbols
a acceleration [m s-2]
D crack height [m]
dp particle diameter [m]
F drag force [N]
f friction factor [-]
g gravitational acceleration [m s-2]
k geometric loss factor [-]
L length [m]
m mass [kg]
p pressure [Pa]
Re Reynolds number [-]
up particle velocity [m s-1]
z height [m]
ε roughness [m]
λ wavelength [nm]
ρ density [kg m-3]
µ dynamic viscosity [m2 s-1]
ν kinematic viscosity [kg m-1 s-1]
σ standard deviation [-]
-
Measurement of particle size distributions in Diesel emission
gasses
4
Chapter 1
Introduction
Diesel engines are widely used of their low fuel consumption and
durability in manycommercial machines, i.e. passenger vehicles.
Compared to petrol enginesthe exhaust gasses contain less carbon
monoxde (CO) and hydrocarbons (HC), far morenitro oxide (NOx) and
particulate matter (PM). Particulate matter is sometimes visible
asblack smoke.For successful application of the Diesel engine in
the future, research of filtering techniquesfor the automotive
sector is called far to satisfy (future) legislation. Calsonic UK
Newcastleand the University of technology Eindhoven started a
project evaluating the applicability of alaser diffracto meter for
measuring the particle size distribution of a Diesel engine.This is
the topic of this project. In this report first the various methods
of filtering arediscussed. After this the measurement technique for
the determination of the particle sizedistribution is explained in
Chapter 3. In Chaper 4 follow the experiments of firstly
solidparticles with a familiar size and secondly measurements with
smoke. Than in Chapter 5 isspoken about the actual experiment with
the Diesel emission gasses. This rapport will endwith some
conclusions and recommendations for further research.
-
Measurement of particle size distributions in Diesel emission
gasses
5
Chapter 2
Particle size distributions in Diesel Engines
This Chapter describes the importance of particle sizes in
Diesel engines and the reasons forresearching this subject. Then
the possible options of measurement techniques which
areinvestigated (in the past) are discussed. Also the Rotating
Particle Separator (RPS) comesup for discussion. This is the direct
cause of this project, because a robust measuringtechnique has to
be developed for the aerosol sampling in this RPS. The goal of
thisresearch is to decide in which direction future development
needs to be taken.
2.1 Diesel engines
Lately environmental and health risks of particulate emission
with respect to their particlesize and distribution has become more
important. The European regulations for the emissionof Hydrocarbons
and NitroOxides in 2005 are three times as strict as they were in
1992 andfor CarbonDioxide even more than five times [3] (Table
1).
Tier Year HC + NOX NOX CO PM
Euro I 1992 0.97 - 2.72 0.14
Euro II – IDI 1996 0.70 - 1.0 0.08
Euro II – DI 1999 0.90 - 1.0 0.10
Euro III 2000 0.56 0.50 0.64 0.05
Euro IV 2005 0.30 0.25 0.50 0.025
Table 1. EU emission standards for Diesel cars, g/kg
-
Measurement of particle size distributions in Diesel emission
gasses
6
Faced with this new legislation manufacturers have to re-examine
their available technologyto meet this challenge. Up till now
regulation has focussed on the emission in terms ofweight, but it’s
expected that attention will shift to the size distribution.Former
measurements indicate that in case of Diesel engine emission, the
size of theparticles is characterised by a bi-nominal distribution.
This distribution has one large peakaround 10 µm as a consequence
of mechanical processes (f.e. agglomeration) and a smallerpeak due
to physical influences (phase change, nucleation) below 1 µm. the
last peak evencan’t be rejected because of the sensitivity of the
human lungs to particles in this range.Due to this new legislations
and the danger for the lungs, also these small sizes have
tofiltered out of the emissions. One of the main issues of this
project is to precise the size ofthis small peak for further
investigation of the filtering techniques
2.2 options for separation techniques
During the last years the following separation techniques have
been developed to reduce theparticulate emission. In principle
these methods are used for different purposes, but withsome further
research and re-design of the familiar plans they probably can be
used inautomotive systems [3]:
-1. impact traps-2. cyclones-3. rotating particle separator-4.
electrostatic filters-5. cracking
Impact traps in combination with regeneration is the only
technique which has matured to thepoint that commercial
introduction is imminent. These filters can catch large particles
due tointerception and very small particles can migrate to the
filter surface by diffusion. Theproblem is that between these areas
the filter is less efficient and probably this is just aroundthe
small peak in the particle size distribution of Diesel emission. A
second thing is theregeneration which is also problem.For a
continuous removal of particles with no moving parts a cyclone is
very effective. Butthe range of particles which the cyclone can
filter has a lower limit of 5 µm, so the interestingsecond peak is
not measurable.The Rotating Particle Separator addresses this
problem by including a rotating cone withvery small channels of 1mm
(figure 1). By reducing the distance between particle and wallthe
cut of size can be reduced to 0.2 µm, which might be sufficient for
Diesel engines.
-
Measurement of particle size distributions in Diesel emission
gasses
7
Figure 1. Rotating Particle Separator
Electrostatic filters perform well for large scale combustion,
but due to therelatively high investments in a high voltage
generator the filter does not scaleeconomically.Thermal Cracking
uses an additional oxidation process to convert the particles
intoH2O and CO2. The efficiency is high, because the conversion
takes place on a molecularlevel. The drawback is the required
burner, what complicates the designand integration of these systems
in a vehicle considerably.When the advantages and the disadvantages
of the different techniques will take intoconsideration a RPS based
system, with the possibility to measure the small peak around1um,
is a possible solution for filtering in small applications. It is
conceivable that also thismethod does not satisfy future
regulations, so the growth of ultra fine particles byagglomeration
to a size which can be filtered is also an important research
subject. If thisproves to be impossible thermal cracking is the
only solution for these very small particles.
-
Measurement of particle size distributions in Diesel emission
gasses
8
Chapter 3
The measurement technique
In this Chapter a possible technique for measuring the particle
size is discussed. Themachine which is used during the measurements
was a Malvern Mastersizer X. Besides thesize of the particles this
machine can also measure the particle concentration and
thedistribution. The Mastersizer X is an optical measurement unit,
which is based on lightscattering. With the method the size
structure of a material phase in another can bemeasured. The only
qualification of the technique is that each phase must distinct
opticallyfrom the other and the medium must be transperant to the
laser wavelength. This means, inpractice, that the refractive index
of the material has to be different from the medium in whichit is
supported. Some advantages are that the method is precise, it is
fast and there is nocalibration required, because the instrument is
based on fundamental physical principles.In the first section of
this chapter the operating principles of the Mastersizer X will
beexplained, after which something is said about the particle size
distributions. At last theexperimental parameters come to
order.
3.1 operating principles
The Mastersizer X is based on the principle of laser ensemble
light scattering [4,5]. It falls inthe category of non imagine
optical systems due to the fact that sizing is accomplishedwithout
forming an image of the particle onto a detector.The optical
configuration which is used is the conventional Fourier optics.
There is also asecond, more accurate configuration, the reverse
optics configuration, but this method canonly be used for particles
dispersed in liquid suspension. So this is not useful
formeasurements on diesel emissions.The conventional Fourier Optics
configuration is shown diagrammatically in figure 2. With
thismethod particles can be measured in a range from 0.5 µ to 600
µm. The light from a lowpower Helium-Neon laser, with a wavelength
of 624 nm, is used to form a collimated andmonochromatic beam of
light. The beam of light (analyser beam) will scatter when it
meetsthe particles in the sample area. The particles are introduced
to the analyser beam by directspraying through the measurement area
or with the help of a cell or pipe of glass, which goesalong the
laser beam. This is why a laser is chosen with a wavelength of 624
nm: in this casethe refraction despite of most sorts of glass is,
especially with large particles, negligible.
-
Measurement of particle size distributions in Diesel emission
gasses
9
Figure 2. Conventional Fourier Optics
The light scattered by the particles and the unscattered
remainder are incident on a receiverlens. This lens is the Fourier
Transform Lens and forms the far field diffraction pattern of
thescattered light at it’s focal plane. Here a custom designed
detector, in the form of series ofangular sectors, gathers the
scattered light over a range of solid angles of scatter.On the
detector the unscattered light is brought to focus and passes
through a small aperturein the detector out of the optical system.
With the total laser power passing out of the systemin this way the
volume concentration can be determined.The Fourier Transform Lens
has the useful property that wherever the particle is in
theanalyser beam its defraction pattern is stationary and centred
on the detector. This is shownis figure 3. Thanks to this property
it does not matter that a particle is moving through thesample
area. The diffraction area stays stationary. It also does not
matter where in theanalyser beam the particle passes. The pattern
stays constant at any lens distance. Onlyhigh particle velocities
(in comparison with the measuring velocity) can influence this
pattern,but because the lens transformation is optical an thus
fast, there are no sample velocitieshigh enough to cause a
deviation from this property.
Figure 3. Properties of the Fourier Transform Lens
The scattered light what falls on the detector is the sum of all
individual patterns. So thesystem measures an integral scattering
pattern form all the particles in the beam. In a typicalmeasurement
the number of particles needed for an adequate measurement is
100-1000depending on their size. Further can 1 single measurement
cause statistical significanceproblems, because the cross section
of the material is to small. To avoid this problems the
-
Measurement of particle size distributions in Diesel emission
gasses
10
method works with a time averaged observation, which gives a
more representative samplingof the bulk material.An unique light
intensity characteristic is produced by the particle when it
scatters the light.So the detector measures a peak at a favoured
scattering area which is related to thediameter of the particle. In
figure 4 is shown that small particles have peak energies in
largeangles of scatter and vice versa.
Figure 4. large against small particle angles
3.2 particle size distributions
In this section facts about the particle size distribution,
which are important for theinterpretation of the experiment are
discussed. The most important point to remember ininterpreting
results is that the fundamental size distribution derived with the
Mastersizer X isvolume based. This means that if there is a
percentage of particles with a certain diameter,the volume of these
particles are that percentage of the total volume. So this method
will notgive the number of particles with that diameter. This means
that the larger particles willpreponderate. In figure 5 is shown
what the difference can be between the two distributions.The
results in this figure are from the same experiment.
Figure 5. volume against number distribution
Another point is that the distribution is expressed in volume
equivalent spheres. In this caseall non-spherical particles will be
transformed during the analysis. This has to be done toavoid that
the height of a cylindrical particle will be seen as the diameter
of the particle andthat the particle according to the analysis is a
lot bigger than it actually is.
-
Measurement of particle size distributions in Diesel emission
gasses
11
At last the scattering of large particles is almost independent
of the optical properties of thematerial and is caused by the
diffraction of light around the particles. Light that couples
intothe particle is absorbed in all common cases and can be
ignored. When the particles aresmaller the refractive index
dependence becomes more significant due to the fact that atsuch
small sizes the light coupled into a particle in not completely
attenuated and canemerge as a refracted ray. For this additional
component a scattering model is required. TheMastersizer X uses the
Fraunhofer theory [Appendix 3].
3.3 experimental parameters
For a correct measurement with the demanded accuracy three
parameters have to becorrect. The first is the instrumental range.
When the distribution is roughly known, the lenswith to correct
significance can be used, otherwise two measurements are
necessary.Another point is the set-up presentation. This means that
data about material andsuspension, like the refractive index have
to be known. At last the analysis model isimportant. There are two
models available in the Mastersizer X software: the polydisperseand
the monodispers model. There are two differences between the two
models. First is theaccuracy [2]:
iMonodispers: the particles have homogeneous physical properties
and a log-normaldiffraction, with a standard deviation of σ ≤
1.2.iPolydispers: the particles have homogeneous physical
properties and a bi-normaldiffraction, with a standard deviation of
σ ≥ 1.2.
Another difference is that in the monodispers model analysis
only the most significant peak isdetected. This model is only
useful if the result falls in a familiar one narrow mode.
Thepolydispers model can measure more peaks in a larger area, but
this is at the cost of thestandard deviation. In the emission
gasses of Diesel are two peaks expected: one largerpeak at 10 µm
and a more interesting smaller peak around 1 µm. Because of the
large peakthe more accurate monodispers model is unusable, so the
polydispers model will be used.
-
Measurement of particle size distributions in Diesel emission
gasses
12
Chapter 4
Spheriglass and smoke experiments
An important issue is the reliability of the results of the
measurements with the Dieselemission gasses. The Mastersizer X has
to measure accurate and the obtained data has tobe analysed
correct. Therefore the machine tested with a familiar distribution
of solidSpheriglass particles. This will be discussed is section
4.3. Also it is necessary to verify themachine can measure a gas
like a Diesel emission. For this reason first an experiment isdone
with a smoke sample. The experiment and the results are in section
4.4. Thesemeasurements acquire a compatible fan, which can
circulate the air. Therefore somecalculations has to be made to
find the required power of the fan (4.2). These calculationsare
based on the test rig, which shall be used in the final Diesel
emission experiment, so thischapter will start with an explanation
about this rig.
4.1 The Diesel emission test rig
In this section the test rig for the Diesel experiment. This is
shown in figure 6. The emissiongasses from the car are flown
through the tube to the T-part. The stream here will be dividedin
two streams. One goes to the Mastersizer, the other one is blown
away. The function ofthis second stream is to aviod an over
pressure in the system de to the power of the engine.After the
Mastersizer a fan is placed, which will drive the gasses through
the laser beam inthe measuring machine. After the fan also these
gasses can be removed.
Figure 6. Diesel engine test rig
A B
C D
A CarB T-partC MastersizerD Fan
-
Measurement of particle size distributions in Diesel emission
gasses
13
4.2 Pressure calculations
The fan that shall be used has to create a pressure difference
what can circulate theparticles. To calculate this pressure a flow
velocity is needed, which is 1 m/s in themeasurements. To create a
flow which carries on the particles, the flow velocity has to
besignificantly higher than the particle velocity, otherwise the
gravitational forces on theparticles influence the flow. So first a
rough calculation will be done, to find the order of theparticle
velocity. To calculate the particle velocity Newtons law (eq. 4.1 )
can be applied toone particle.
In this formula m is the mass of a particle, a the acceleration
of a particle, what is equal tothe gravitational acceleration g. F
is the drag force. For one particle this is:
With νf the kinematic viscosity, dp the diameter and up the
velocity both of one particle.Formula 4.1 can be written as:
Here m is written as the volume of a particle multiplied with
the density, which is settled withthe kinematic viscosity ν. Now
this becomes the dynamic viscosity µ.The viscosity of air is
2.16*10-5 and g is 9.81 so out of 4.3 follows up=151dp. The
meandiameter is in order of micrometers, so as a conclusion can be
said that 1 m/s for the flowvelocity is much higher than the
particle velocity and can be used in the pressure calculationsand
the experiments.For the calculation of the pressure the Bernouilli
equation will be used [1]. When frictionlosses and losses due to
the geometry of the canals are added, the equation becomes:
In equation 4.4 p is the pressure, ρ the density, z the height,
v the velocity, f the frictionfactor, L the pipe length, D the
diameter of the pipe and k the geometric loss factor. thesubscript
1 is for the place at the beginning of the pipe just after the car
and subscript 2 atthe end after the fan. In this system v1 = v2, z1
= z2 so 4.4 reduces to:
amF ⋅=
ppf udF ⋅⋅⋅=23 νπ
ppfp udgd ⋅⋅⋅=⋅23 3
6µπ
π
g
vk
g
v
D
fLz
g
v
g
pz
g
v
g
p
2222
22
2
222
1
211 ⋅∑+⋅∑+++=++
ρρ
22
22 vk
v
D
fLp
ρρ⋅∑+⋅∑=∆
4.1
4.2
4.3
4.4
4.5
-
Measurement of particle size distributions in Diesel emission
gasses
14
In appendix 1 a table is given with geometric loss factors k.
For the different parts of the testrig the following k values can
be found:
-T part 1.9-Inlet 1.0-inlet fan 1.0-----------------------total
3.9
For the calculation of the friction factor f the Reynolds number
is needed:
The viscosity has a value of 2.16*10-5 and v is 1 m/s (see
above). In the test rig rubber tubeswith a diameter of 25mm are
used and for the density of emission gasses an approximationis made
by comparing it with air. (ρ = 1.06 kg/m3). The Reynolds number
then becomes1.47*103. Now the friction factor can be found in
figure 7. For ε / D a value of 5*10-7 is used,so a friction factor
of 0,054 will follow.
Figure 7. Moody diagram
When the total length of the pipes is 5 meters the total
pressure fall can be calculated withequation 5 This will be 6.5 Pa.
To cause this pressure fall, a fan of the company ABC.A typeCK 125C
exp. is used.
4.3 The Spheriglass experiment
For the inspection of the Mastersizer X an experiment has been
done with Spheriglassparticles. This is a material what is used in
road building for the sparkling of the white marks.From former
experiments with a different measuring technique, the particle size
distribution isknown. This method can be used for solid particles
with sizes which are not smaller than 5µm, but for emission gasses
it is not suitable.
µρvD
=Re 4.6
-
Measurement of particle size distributions in Diesel emission
gasses
15
4.3.1 the test rigIn the experiment a flow of solid particles
has to cross the laser beam. To prevent that as aconsequence of
scattering the flow is not measureble, a pipe of glass is used to
lead theparticles through the laser beam. In the Diesel emission
experiment this technique isn’tpossible because the refraction of
the glass is large in comparison with the small diameter ofthe
particles in the second peak (Chapter 3.3). Thanks to the larger
diameters of theShperiglass particles, it is not a problem in this
experiment.In figure 8 the test rig is shown. Due to the suction
force of the fan an air circuit arises, whatleads the particles
around. In this experiment a problem was that the fan which
wascalculated for emission gasses was not capable moving the
particles, so the fan wasreplaced by a vacuum cleaner.
Figure 8. test rig of the Spheriglass experiment
4.3.2 the resultsDuring the experiment the following results
have been measured (figure 9, app. 2). If thesevalues are compared
with the results of former measurements of Van Beek, the top of
bothgraphs are comparable and lie around 60 µm.
Figure 9. The Mastersizer experiment Van Beeks experiment
A difference is that the particle size distribution in this
experiment is wider. A possibleexplanation is that the Mastersizer
has less measuring points so the distance between these
B A
C
D
A FanB MastersizerC pipe of glassD hose of rubber
Particle Diameter (µm.)
Volume %
0
10
20
30
0
10
20
30
40
50
60
70
80
90
100
1.0 10.0 100.0 1000.0
Particle Diameter (µm.)
Volume %
0
10
20
30
0
10
20
30
40
50
60
70
80
90
100
1.0 10.0 100.0 1000.0
-
Measurement of particle size distributions in Diesel emission
gasses
16
points is more than at Van Beeks experiment. Due to this also
the volumetric percentage isnot the same. Despite of these
differences the conclusion can be made that the MastersizerX is
compatible and the experimental parameters (Section 3.3) are
correct.
4.4 smoke experiment
In this experiment smoke of tobacco is used to control if the
Mastersizer is capable tomeasure the small particles in this gas.
During the test the smoke was blown directly throughthe laser beam,
without using any fan or circuit. This is done to avoid the pipe of
glass, whichinfluences the results (4.3.1). The problem that was
risen in the Spheriglass experiment (thescattering of the gas
before the laser beam was crossed) is not a problem here,
becausesmoke can be blown directly through the Mastersizer. This is
not possible with solid particles,because of gravity influences on
these particlesIn figure 10 (and app. 2) the results are shown. A
large peak around 15 µm can be noticed,but also the second, for the
RPS more interesting small peak in measured. This peak liesabout
0.75 µm. As a conclusion can be said that the Mastersizer is
capable to measuresmoke and has to be able to measure emission
gasses as well
Figure 10. Results smoke experiment
4.5 explanation of possible problems
During the measurements some problems can arise, which can cause
a difference betweenthe real diffraction pattern and the measured
one. This paragraph gives an explanation forthe difference.First
the pipe of glass causes a deviation. The refractive index of glass
can influence thescattering of the light and for some wavelengths
of light, the glass will not let the lightthrough. So the glass
cannot be thick (up to 2 mm) and has to let through a wavelength
of633 nm (the wavelength of the laser light). Also the pollution of
the pipe can cause a differentrefractive index and through that
deviation.Another point is the transformation of non-spherical
particles to particles with equivalentspheres. For small particles
this is not a problem, but the deviation becomes larger if
theparticles are larger. In emission gasses appear besides small
particles also very large
Particle Diameter (µm.)
Volume %
0
10
20
0
10
20
30
40
50
60
70
80
90
100
0.1 1.0 10.0 100.0 1000.0
-
Measurement of particle size distributions in Diesel emission
gasses
17
particles (soot or ash). The number of these very large
particles is not high but they can havea big influence in a volume
based distribution.Further the light has to cross the gasses. For
the surrounding material a refractive index isgiven to the model.
This is the index for air, the emission gases, which flow through
the pipecan cause a deviation, because the refractive index for
this gas is not exactly the same.At last some small particles can
stay behind in the rubber tube before the laser, as aconsequence of
the roughness of the wall.
-
Measurement of particle size distributions in Diesel emission
gasses
18
Chapter 5
Diesel emission experiment
In this chapter the measurements with Diesel emission gasses are
discussed. The gassesproduced by a car are led along the laser lens
and measured by the Mastersizer. Twoexperiments were done. In the
first experiment the car was only started, without adding morefuel.
In the second one this was actually done. The results are graphed
in figures 11 andAppendix 2
Figure 11. results without extra gas results with extra gas
When gas is added larger particles are measured. A logical
explanation is that moreemission gasses with lager particles are
produced , because not all the fuel, which is addedcan be burned.
An important result is that the interesting small peak in can be
measured.This peak lies around 0.8 µm (figure 11). In the right
figure this peak falls away for the peakwith lager particles. So
the Mastersizer meets the demands, if the larger particle peaks
canbe suppressed.
Particle Diameter (µm.)
Volume %
0
10
20
0
10
20
30
40
50
60
70
80
90
100
0.1 1.0 10.0 100.0
Particle Diameter (µm.)
Volume %
0
10
20
0
10
20
30
40
50
60
70
80
90
100
0.1 1.0 10.0 100.0 1000.0
-
Measurement of particle size distributions in Diesel emission
gasses
19
Chapter 6
Conclusion and recommendation
One of the most important power sources this time is the Diesel
engine. This is because oftheir low fuel consumption and their
durability. A problem is the emission gasses, which haveto become
cleaner in the future to satisfy the environmental legislations.
For this reason thefiltering techniques become more important and a
lot of research is done on this subject.For further research on
filtering techniques first a measurement technique for the
particlesize distribution has to be developed. This was done in
this project, to examine whether itwas possible to measure this
size distribution with a laser diffracto meter, a
MelvernMastersizer X.After positive results in testing the
Mastersizer with a familiar size distribution of
Spheriglassparticles and smoke, measurements were done with
emission gasses. As a conclusion canbe said that the laser
diffracto meter can be used for the regulation of particle
sizedistributions as can be seen in the results of Chapter 4 and
5.
On this subject no recommendations have to be made. With this
measurement technique theparticle size distribution can be
measured. A better alternative for further research are
thefiltering techniques itself, for example utilising the Rotating
Particle Separator automotive.
-
Measurement of particle size distributions in Diesel emission
gasses
20
Literature
1. William S. Janna, Design of fluid thermal systems, 1993, ISBN
0-534-93373-4
2. B.A.R. van Peer, seeding for LDA in het bijzonder voor
toepassing inverbrandingsmotoren, rapportnummer 97017,1997
3. H.P.van Kemenade, W.Sampers, Proposal for an investigation on
Diesel particulateemissions and after treatment technology,
1999
4. Malvern Instruments, Applying advanced particle science in
instrumental & research,Instrumental manual, Manual number
0054
5. Malvern Instruments, Applying advanced particle science in
instrumental & research,Windows sizer reference manual, manual
number 0073
6. Douglas C. Giancoli, Natuurkunde voor wetenschap en techniek
deel II, 1993, ISBN 9062339077
-
Measurement of particle size distributions in Diesel emission
gasses
21
Appendix 1
Geometric loss factors
-
Measurement of particle size distributions in Diesel emission
gasses
22
-
Measurement of particle size distributions in Diesel emission
gasses
23
Appendix 2
The numerical results of the measurements
Spheriglass experiment (figure 9)
Particle Volumesize percentage
0.2 0.080.48 0.270.59 0.420.71 0.480.86 0.461.04 0.37 1.26
0.241.52 0.131.84 02.23 02.7 03.27 03.96 04.79 05.79 0 7.01 08.48
010.27 012.43 015.05 018.21 0 22.04 0.0826.68 0.8632.29 3.7939.08
11.2847.3 20.0957.25 22.7969.3 19.1583.87 12.57101.52 5.8122.87
1.15148.72 0
-
Measurement of particle size distributions in Diesel emission
gasses
24
Smoke experiment (figure 10)
Particle Volumesize percentage
0.20 0.490.48 1.890.59 2.720.71 2.810.86 2.291.04 1.491.26
0.751.52 0.251.84 02.23 02.70 03.27 03.96 04.79 0.205.79 1.367.01
3.908.48 8.0210.27 12.6312.43 15.7115.05 15.9818.21 13.5522.04
9.2026.68 4.8132.29 1.7539.08 0.1847.30 057.25 069.30 083.87
0101.52 0122.87 0148.72 0
-
Measurement of particle size distributions in Diesel emission
gasses
25
Diesel emission experiment without adding extra fuel (figure
11)
Particle VolumeSize percentage
0.20 0.770.48 3.710.59 5.960.71 7.210.86 7.431.04 6.881.26
5.951.52 5.051.84 4.432.23 4.062.70 3.773.27 3.443.96 2.914.79
2.285.79 1.797.01 1.558.48 1.6010.27 1.8812.43 2.3515.05 2.8718.21
3.3322.04 3.6126.68 3.6232.29 3.3439.08 2.9447.30 2.5657.25
2.1369.30 1.6683.87 0.92101.52 0122.87 0148.72 0
-
Measurement of particle size distributions in Diesel emission
gasses
26
Diesel emission experiment with adding extra fuel (figure
11)
Particle Volumesize percentage
0.20 0.190.48 0.180.59 0.100.71 00.86 01.04 0.041.26 0.401.52
0.901.84 1.492.23 2.292.70 3.573.27 5.533.96 7.354.79 8.205.79
8.047.01 7.478.48 6.8910.27 6.1812.43 5.4315.05 4.8318.21 4.2722.04
3.7126.68 3.1632.29 2.6139.08 2.1247.30 1.7157.25 1.4169.30
1.2183.87 1.35101.52 1.78122.87 2.69148.73 4.88
-
Measurement of particle size distributions in Diesel emission
gasses
27
Appendix 3
Fraunhofer theory
Suppose a monochramatic source of light which falls through a
crack [figure A1]. The widthof the crack is D and the light
consists of parallel beam. Is the screen placed on a
infinitedistance the diffraction is called Fraunhofer, instead of
when the screen is placed nearby thecrack, then it’s Fresnel
diffraction [6].
Figure A1 monogramatic parallel beam through crack
Because the screen is at an infinite distance the beam stays
parallel until an arbitrary pointon the screen. In figure A1.a a
beam goes straight through the crack. All the streaks of lighthave
the same phase so a clear spot arises. When the beam falls through
the crack under anangle the streaks above has a longer optical path
to travel than the lowest streak. In figureA1.b the difference
between the middle and the lowest streak is λ/2, so the phase
differenceis 180°. Therefor they interfere destructive. this also
counts for the difference between themiddle and the highest streak.
The consequence of this interference in pairs is that no
lightreaches the screen under this angle. This angle is:
D/sin λθ = A.1.
-
Measurement of particle size distributions in Diesel emission
gasses
28
When the difference in distance is larger, for example 1,5 λ,
the destructive interference inpairs arises between the lowest
streak and the streak at one third, because this difference isλ/2
[A1.c]. The light from the highest streak do arise at the screen,
only the spot is muchweaker than the spot in the middle. The
following light pattern arises [figure A2]
Figure A2 Fraunhofer light pattern