design sr analysis catalytic converter -...

CHAPTER- 3

DESIGN sr ANALYSIS OF CATALYTIC

CONVERTER

57

Design & Analysis of Catalytic Converter

3.1 Design Aspects of Catalytic Converter.

The design of catalytic converters for 2-stroke engines requires a system

approach because of trade-offs involved in arriving at the optimum design.

The following requirements must be satisfied in designing the catalytic

converters:

01. Should provide enough surface area to meet conversion

efficiency requirement for both HC + N02 and CO.

02. Power loss due to backpressure created by the converter should

be minimized.

03. Should not interfere with engine operation.

04. Must meet durability requirements.

05. Heat generated by the catalytic reactions must not raise safety

converters.

06. Must fit in the available space.

07. Should be affordable.

3.2 Robust Design and Analysis of Substrate.

Robust design is a methodology that addresses product quality issues early in

the design cycle. The goal of robust design is to deliver customer

expectations at profitable cost regardless of customer usage, degradation

over product life and variation in manufacturing, suppliers, distribution,

delivery and installatton Since randomness and scatter is a part of reality

5X

everywhere, probabilistic design techniques are necessary to engineer quality

into designs Traditional deterministic approaches account for uncertainties

through the use of empirical safety factors. The safety factors are derived

based on past experience; they do not guarantee safety or satisfactory

performance and do not provide sufficient information to achieve optimal use

of available resources. The probabilistic design process has not been widely

used because it has been intimidating and tedious due to its complexity. In

this research effort, probabilistic modeling of manufacturing and material

variations for a catalytic converter substrate was considered. Typical shapes

of catalytic converter substrates are shown in Figure 3.1.

Figure 3.1 -Typical shapes of catalytic converter substrates

The substrate used in this study has a cylindrical cross section and is

enclosed in a cylindncal steel cover If the substrate is not a perfect cylinder

59

the steel cover applies a non-uniform pressure along the circumference.

Assuming that the maximum diameter of the substrate is Dmax and the

minimum diameter is Dm;n, we can characterize the variation in circularity or

roundness d with their difference

Due to manufacturing variations d is considered a random input variable.

Figure 3.1 Typical shapes of catalytic converter substrates In this study, it was

assumed that a Gaussian distribution with mean value md and standard

variation sd characterizes the variation in circularity d. Due to both material

and manufacturing variation the ultimate shear stress 'tult exhibits

randomness. A pure shear test was performed on several substrates to

determine the ultimate shear stress variation. The mean value of ultimate

shear stress is m'tult = 0.2868 MPa (41.6 psi), the standard deviation is S'tu1t =

0.01724 MPa (2.5 psi), the minimum value of the sample was ('tu1t) min = 0.255

MPa (37.0 psi) and the maximum value of the sample was ('tu1t) max = 0.31

MPa (45.0 psi).

The objective of this study is to identify the supplier specification (sd) max

(maximum standard deviation of variation in circularity d) in order to achieve a

robust design of a des1red sigma quality level.

60

The Parametric Deterministic FEA Model

A parametric finite element model of the substrate is shown in Figure 3.2 and

was developed considering the following assumptions:

1. The material of the honeycomb ceramic substrate is isotropic,

linear elastic and the behavior is within small deflection linear

theory limits. The modulus of elasticity is E = 4800 MPa and

Poisson's ratio is 11 = 0.25.

2. Plain Strain Analysis is sufficient to accurately predict the

maximum shear stress.

3. There is no temperature effect on the material properties.

4. The geometry, loading and behavior are symmetric about the

horizontal and vertical axes, thus a quarter symmetry model can

be used.

5. The angle between the maximum and minimum diameter is 90°.

6. The geometry of the FEA model can be represented by the first

quadrant of an ellipsoid with the horizontal lower edge length

equal to F/2 + d/4, and the vertical left edge length equal to F/2 -

d/4, where F = 105.0 mm.

7. The maximum pressure Pmax (in MPa) is a function of .8 (in mm)

and can be computed by the following equation:

Pmax(o) = a5,)5 + a4 84 + a3 ,i3 + a2 &2 + a1 <i1 +aO

where:

ao = -0.8155

a1 = 5.5840

a2 = -8.3864

a3 = 4.0882

a4 = -0.0694

a5 = 0.0004

61

Figure 3.3 shows a plot of this function (the maximum pressure

Pmax versus 6) and the data-points used for the curve fitting.

8. As shown in figure 3.2 the maximum pressure is applied at the

point of maximum diameter and varies sinusoidal to zero at the

point of minimum diameter:

P(fl) = Pmax *COS (8)

Figure 3.2 -Finite Model and Loading of the Substrate

·r- ----r--

'

'

'

0 ''

J v 1/

v _.,.. /

..,...,.,.. v -I 1 1 :! 1 1 I ,J 1 -. I 1, I (

s l11Hlll

Figure 3.3 Maximum Pressure P m" versus I>

62

The input parameters of the model are variation in circularity d and ultimate

shear stress tu 11• For any combination of the input parameters the solution to

the parametric model can compute the maximum shear stress tmax- Figure 3.4

shows a typical distribution of the magnitude of the displacements and figure

3_5 shows a typical maximum shear stress distribution_

63

1\N

Figure 3.4 Typical Displacement Distribution

1\N or;-.· • n"4

'·.:. }I

Figure 3.5 Typical Maximum Shear Stress Distribution

We define a performance function G as the difference between ultimate shear

stress and the maximum computed stress or G = tun -'tmax· If G remains positive

at all times we will have a safe design. The performance function G is

considered as the output variable and is a function of the input variables d and

3.3 The Probabilistic FEA Model

Uncertainty in the input parameters of the FEA model can be introduced by

assuming certain randomness in the input parameters. In this study, it was

assumed that a Gaussian distribution with mean value md and standard

variation sd characterizes the variation in circularity d.

The mean value md was considered uncontrollable or a noise parameter with

a constant value of md = 1.05 mm. The standard variation sd was considered

as a controllable parameter and it was declared as an optimization design

variable. For various values of the standard variation sd one may obtain a

distribution for d. Figure 3.6 shows the Probability Distribution of the input

variable 8 with a standard variation sd = 0.01 mm.

>. u

Histogram of Circularity Var tatwn,;; ·1'-',----,---,---,--..... -.---.---,---,

;; ,. lf----1--+--5-~

u. ''lif--+--+--,, -~ ~ Jl"··ll---t--+-;;; a:

"' Figure 3.6. Probability Distribution of the Input Variable 8

It was also assumed that truncated Gaussian distribution characterizes the

ultimate shear stress of the substrate material. The mean value of ultimate

65

shear stress mt .. 11 = 0.2868 MPa, the standard deviation S<ult = 0.01724 MPa,

and the range 0.255- 0 310 MPa were determined experimentally. Figure 3.7

shows the Probability Distribution of the input variable Stu11 with these values.

Ht"it(HJI :.m of Ult1matf! Shear Stress t\111

•,(·,------,-----, ~-~

llll11nate ShP.ar St1ess '••ll MPa

Figure 3. 7 - Probability Distribution of the Input Variable <ult

Since the two input variables are random the performance function G (where

G = (<u11 - (<u11) max) exhibits randomness and is considered the output variable.

H1stogram uf Pedormance Function G for a6 = 0.01mm

• .!,-1 c------+---+----1--t--

., ' ,,.-,, ., ,-, F·

Perfor manc.e Func.t!Oil G MPa

Figure 3.8 Probability Distribution of the Performance

Function G for Standard Variation 0, = 0.01 mm.

66

H~-:>IOUI am of PelfOIITii'lllCP. Funct1011 G foro~= 0.05 mm

1_1 '.

"' ', f ~--11----11--~t--

gLc-,~~.,"7,,~~- 0(J5 •J1 (I~' Perf01 manca- Function G MPa

Figure 3.9 Probability Distribution of the Performance

Function G for Standard Variation cr, = 0.05 mm.

Monte Carlo and the Central Composite Design response surface sampling

techniques were implemented in determining the response distribution of the

output variable for various values of the standard variation cr,. The probability

distribution of the performance function G for standard variation cr, = 0.01 mm

is shown in Figure 3.8. The mean value of the performance function is m0 =

0.08312 MPa and the standard deviation is Sc; = 0.01521 MPA. One may

observe that the entire distribution of the performance function G remains on

the positive side, indicating that for cro = 0.01 the maximum shear stress does

not exceed the ultimate shear stress. In this case the probability that the

performance function G is less than zero is 0 %, P[G<O] = 0%. Figure 3.9

shows the probability distribution of the performance funct1on G for standard

variation n,, = 0.05 mm The mean value of the performance function is mG =

0.07849 MPa and the standard deviation is sG = 0.04026 MPa. One may

67

observe that part of the distribution of the performance function G remains on

the positive side indicating that for cr0 = 0.05 the maximum shear stress

exceeds the ultimate shear stress. The area to the left of the zero (red line)

indicates the probability of failure. In this case the probability that the

performance function G is less than zero is 4.24 %, P[ G< 0]:: 4.24%.

3.4 Further Analysis Of Substrate

If the catalyst coating is the same, the first three design considerations

depend on substrate parameters. Higher cell density is desirable from a

conversion efficiency point of view. However the pressure drop, i.e., power

loss, will be greater with higher cell density Table-3.1, show the values of

geometric surface area and flow resistance parameters for different cell

density. It was found that 100 cells/in2 is an optimum cell density for

motorcycles to meet surface area and pressure and requirements.

Table 3.1: Geometric Surface Area and Flow Resistance Parameters.

I Cpsi !Wall thickness GSA RF i I

I Cell/in2 ! M2/m3 1/m2 :Mm

I

I

I

[100 , ______

i 0.30 1386 7.34 I ' '200 ! 0.20 1975 14.23

' I

350 ' 0.14 2643 23.79

'

400 : 0.17 2740 30.74 ! I

6X

Once the cell density is chosen, the next step in the design process is the

selection of the size and aspect ratio of the ceramic substrate. Typically the

volume of the ceramic substrate for motorcycle application ranges from 75-

200 ml depending on the size of the engine and available space. If the

substrate volume and cell density is held constant, the aspect ratio of a

cylindrical converter affects the pressure drop in the following manner.

Pressure drop a L/02

Where, L and D are substrate length and diameter, respectively. The aspect

ratio is also important in determining the mat/substrate contact area, which is

important in determining the holding force provided by the mat for cylindrical

converters, at constant volume. The mat contact area decreases with

increase in diameter of the substrate. Depending on the available space and

location of the converter, the aspect ratio (LID) of motorcycle converters can

vary from 0.8 to 2.5.

3.5 Durability

The durability of a catalytic converter can be limited by either the catalyst

durability or the phys1cal durability of the converter package. Factors that

determine catalyst durability are

01 Temperatures w1th1n the converter

6'1

02. Impurities in the fuel and lubricating oil.

03. Wash-coat composition and method of application composition and

distribution of precious metals within the wash-coat

04. Geometric surface area of the converter.

The temperatures within the converter of a 2-stroke engine can reach 1050°C

due to exothermic heat generated by the oxidation of hydrocarbons and CO.

State-of-the-art catalysts can withstand this temperature for short duration

(@25 hours). The catalyst loses its activity due to loss of surface area of the

wash-coat Sometimes chemical interactions between the precious metals

can also lower the activity of the catalyst Therefore, it is necessary to

minimize temperature excursions within the converter to maintain longevity of

the catalyst coating. If the temperatures within the converter are maintained

below 900°C, it is possible to obtain long-term durability in motorcycle

application. This assumes that the impurities in the fuel and lubrication oil do

not deteriorate the catalyst

Physical durability, i.e., the converter's ability to withstand thermal and

mechanical stresses, is an important consideration in determining its life. The

parameters that determine the physical durability of the ceramic catalytic

converter are:

01. H1gh temperature physical properties of the ceramic substrate

02 The magnitude of the holding force provided by the mat

70

03 Mat durability

04. High temperature creep resistance and thermal expansion of the

stainless steel can

Strength, thermal expansion coefficient and elastic properties are the main

physical properties that govern the durability of the ceramic substrate. The

physical properties of the ceramic substrate (EX-22) are given in table 3.2 and

figure 3.1 0. For comparison, the high temperature strength behavior of

metallic substrates is shown in figure 3.11. The biaxial compressive strength

of the ceramic substrate is ten times greater than the radial compressive

stress imposed by the mat (500 KPa), and therefore, the failure of the

substrate due to mechanical loads is not a significant concern. The same is

true with respect to thermal stress failures.

Table 3.2. The physical properties of the ceramic substrate

,.---··· ·-,. Thermal Expansion RT-1000°C (10 -7f'C) Elastic Modulus (GPa)

1000°C 1100°C

I Density (g/cc) I l_l3iaxial Strengtll (f<.i=>_a_l __j

-- --

Substrate

Bare With

Catalyst 5.2 10.4

9.3 13 14.5 18

, I 0~51 0.8 i 6500 i

--- ----

., 7

~ 6 ;; 5

"' c: 4 ~ Ul 3 iii 2 ~

;l 1 Q)

iL 0

Axial direction

~ .--.

..._<:><:> n,<:><:; ""<:;<:;

Temperature (oC)

71

I

___... _._...

' ' ! I

Figure 3.10: High temperature strength behavior of EX-22 ceramic

substrates .

., 800 ~ 700 ;; 600 c, 500 c: ~ 400 Ul 300 .21 200 ·~ 100 ~ 0

0

I

""' I

""-. ' ' """-. ' " I

\ I \. I

~ I

500 1000 1500 Temperature (oC)

Figure 3.11: High temperature strength of metal substrates.

The Magnitude of thermal stresses depends on the temperature gradient

across the substrate. thermal expansion coefficient, elastic modulus and

Poisson's ratio of the coated substrate. Since the size of motorcycle

converters is much smaller than the automotive converters, the temperature

gradients w1ll be lower. Failure of the substrate due to thermal stresses in

motorcycle application is unlikely because ceramic substrates have been

performing reliably for the last 20 years in car application. As mentioned

earlier, the temperatures inside the catalytic converter of a 2-stroke

motorcycle can reach 1 050°C. For satisfactory performance, the high

temperature strength of the substrate must be adequate, As seen from the

data in figure 3.10 and 3.11 the ceramic substrate maintains its strength up to

1200 °C. Whereas the strength of the metallic substrates drops significantly

above 600 °C. Distortion of the cell walls can occur if the high temperature

strength is not adequate. Cell distortion can increase pressure drop across

the convector, which in turn causes an increase in power loss of the vehicle.

Another issue that needs to be addressed is the sudden rise in converter

temperature due to unburned fuel entering the converter. The melting

temperature of the substrate should be high enough to withstand the

temperature rise. The melting temperature of the substrate should be high

enough to withstand the temperature rise. The ceramic substrate (EX-22)

melts at 1465'c which 1s 400'c higher than the maximum allowable

temperature for the catalyst.

3.6 Mat

Because of the large thermal expansion difference between the

substrate and the stainless steel can, a mat is needed to secure the ceramic

substrate in the steel can. The mat must exert enough radial force on the

7.1

substrate to hold it in place in the exhaust enVIronment of a 2-stroke engrne. It

is possible to estimate the forces on the converter due to chassis vibrations

and pressure drop across the converter. The total force, F on the ceramic

substrate in the·direction of gas flow is given by

F=W (V/G) + PA --------- (1)

Where W, is the mass of the coated substrate, V is the maximum

vibrational acceleration experienced by the ceramic substrate, G is the

gravitational acceleration. P is the backpressure, and A is the cross sectional

area of the substrate. If one assumes V as 120 G and P as 30 g/cm2 (30 em

of water). the maximum force on a 55 x 40 mm long substrate is 96 N. The

shear stress at the substrate I mat interface due to this force is 14kPa. The

radial stress. i.e .. holding pressure to be provided the mat can compute p,

computed from the shear stress and the coefficient of friction, ftJ. p =shear stress I j.l------------- (2)

Assuming mas 0.2. the minimum radial pressure to be provided by the

mat is 70 kPa. It will be shown later in the paper that non-intumescent mats

can provide the necessary holding force. Before moving to design aspects of

the stainless steel can. rt is necessary to mention that the forces generated on

the substrate would be much higher if the vibrational frequencies encountered

by the converter coincrdes with the resonance frequency of the conversion.

The vibrational frequencres encountered by the converter can be measured

by mountrng accelerometers on the surface of the steel can.

7-1

3.7 Stainless Steel Can

The steel can is exposed to high temperatures because of the heat

generated within the converter. It is necessary to select steels that have low

thermal expansion coefficient and excellent creep resistance to maintain the

holding force provided by the mat. The holding force can decrease if the steel

shell expands too much or if the steel undergoes permanent deformation.

Ferrite stainless steels 409 and 439 are commonly used for catalytic converter

cans. The thermal expansion of this steel is 30% lower than 300 series

stainless steels. The yield strength of these two alloys at different

temperatures are given in table 3.3. The hoop stress generated in the steel

can depend on the radial pressure provided by the mat, the diameter of the

can and thickness of the can. For circular cans, the hoop stress, a is given by

0 = p d,J 2 t -----------------(3)

Where p is the radial pressure exerted by the mat, d is the diameter of the

substrate and t is the thickness of the steel can. Assuming p= 700 kPa, d=60

mm and t=1.2 mm, the hoop stress in the steel can is 17.5 Mpa. The yield

strength data in table 3.3 indicate that the permanent deformation at 17.5 Mpa

will be negligible up to 800oC for 409 steel and up to 875oC for 439 steel. In

most cases, the motorcycle catalytic converters are mounted inside the

mufflers and therefore. the skin temperature of the converter can reach 700-

800oC 1n 2-stroke veh1cles

75

Table 3.3: Yield Strength of 409 and 439 Steels (Mpa)

Temperature °C 409 Steel 439 Steel ---- --

25 244.7 302.6

204 173.7 245.4

316 169.6 253.0

482 147.5 215.1

538 135.1 179.9

593 111.7 191.0

649 92.4 137.2

704 46.9 82.7

760 29.0 57.9

816 22.1 42.7

871 15.2 30.3

I 927 6.2 17.2

I , ___ --- ----

3.8 Results

The objective of this study is to establish that it is possible to achieve the

following performance targets using catalytic converters on 2-stroke

Motorcycles and scooters.

01. HC+ N02 and CO emission levels below 60% of year 2004 standard on

I DC. ie .. HC + No2 < 0.9 g/km and CO < 1.2 g/km.

02. Less than 5% power loss over the entire speed range

03. Must fit in available space without interfering with normal operation of

the b1ke

04 Meet the des1red on-the road durability.

7h

The progress made to date in achieving the above objectives will be

discussed in this section. Additionally, the results of several new tests that

were conducted to characterize the mat durability are also included in this

paper.

3.9 Emission Results

Engine out emissions from motorcycles and scooters in IDC vary considerably

depending on engine design, engine size, carburetor design, and air/fuel ratio.

The HC + N02 values of different vehicles can range from 3 to 6 g/km and for

CO from 1 to 7 g/km. Depending on engine out emissions, conversion

efficiencies greater than 60% are needed to meet the year 2000 emission

standards. Mass emissions tests were conducted on a 150 cc scooter and a

110 cc motorcycle using a chassis dynamometer. As prescribed, emissions

were collected over six measuring cycles after four warm up cycles. The

variables in the experiments were, cell density, size of the ceramic substrate,

catalyst formation. secondary air injection, catalyst location, carburetor

adjustment and a pre-converter to increase the exhaust gas temperature.

Tables 3.4 and 3.5 provide emission results obtained on these two vehicles

for the 11 o cc motorcycle, eight warm up cycles were used prior to emission

measurement although catalyst light off occurred in the fourth cycle.

Table 3.4: Emission Test results of 150 cc Scooter:

Conversion Efficiency

f.--------Baseline (engine out)

Idling CO at 0.62%

Idling CO at 1.2%

·-----

---+---

--

HC c 0

g/km g/k m

3.07 2. 63 --------

1.1 0. 70

1.1 0. 90

(%) ---- --

HC

-----

--

64

64

Table 3.5: Emission Test results of 110CC motorcycle:

HC (g/km)

----------· Baseline (engine out ) with 100 cpsi, 55 x 40mm 5.65

..

1.69 1------------Catalyst

1----- -- - . - --~-Conversion Efficiency (%) 70

--

-- - -----

co

---

74

66

co (g/km) -

1.2

0.54

55

77

It can be seen from these results that both HC and CO emissions were

reduced significantly using ceramic catalytic converters. The emissions levels

of N02 in 2-stroke engines are generally bellow 0.02 g/km and therefore, is

not a big concern. The emission data in table 4 suggests that carburetor

adjustments in idling condition can affect CO conversion efficiency.

The em1ssion levels of the 110 cc motorcycle could not be reduced below the

year 2000 standard because of higher baselme HC emissions an limited

catalyst volume By mcreasing the size of the catalyst or by secondary air

injection, it should be possible to reduce HC emissions of the motorcycle to

year 2000 standard.

3.10 Power Loss

The power of 2-stroke engines is very sensitive to any modifications in •

the exhaust system. Two-stroke engines do not have valves and therefore,

the backpressure created by the catalytic converter can interfere with engine

operation. Since vehicle power is an important performance parameter, the

installation of a catalytic converter in the exhaust system should ideally have

minimal effect on engine power. Realizing that the vehicle power can vary as

much as 2% due to statistical variations in muffler dimensions, it was decided

that the power loss due to the catalytic converter should be less than 5% over

the speed range of the vehicle.

The vehicle power was measured on a chassis dynamometer for both the 150

cc scooter and the 110 cc motorcycle. The scooter had two catalytic

converters in the exhaust system one small converter closer to the exhaust

port to raise the gas temperature to light off the main catalyst. Note that the

emission results in table-4 were obtained with a similar catalyst arrangement.

The power of the motorcycle was measured with a converter, which had a 55

x 40 mm 100 cell/in 2 substrate Figures 3.12 and 3.13 show the change in

engine power at drfferent speeds due to the catalytic converter. In these

graphs. values above zero represent power gain and values below zero

represent power loss

7'1

~ .. 8 ;: 6 0 a. 4 • .. .!: 2 • • Cl-

.TI~ 0

.!: -2 'V ,..,. 0 .. -4 Cl

t:

"' -6 .t: u -8

• I • •

Speed (km/h)

Figure 3.12: Power loss of 150 cc scooter with ceramic catalytic

converter.

~ 15 . . .. .. "' ;: 0 10 a. "' = 5 m_ =~ W'!... • • I <: 0

"' Cl -5 <:

• ~ 20 40 liP 100 120 1 0

"' .s;:: 0 -10

Speed (km/h)

Figure 3.13: Power loss of 110 cc motorcycle with ceramic catalytic

converter.

For both vehicles the power loss varied with vehicle speed. Note that at

certain speeds the veh1cle power was greater with a catalytic converter. With

the exception of one speed, 40 km/h for the scooter and 85 km/h for

motorcycle. the power loss target of 5% was met for both vehicles. These

results suggest that by JUdicious selection of catalyst location cell density size

xo

an aspect ratio of the substrate ceramic catalytic converters can be installed

with minimal effect on vehicle power.

3.11 Durability Of Ceramic Converter

Several aspects of the catalytic converter need to be considered to achieve

the required durability. The first one is the deterioration in conversion

efficiency. Conversion efficiency depends on factors such as the maximum

temperature within the converter, the impurities in fuel and lubricating oil, and

geometric surface area. In addition to these, the change in engine-out

emissions due to deterioration of carburetor and spark plug performances will

determine the overall effectiveness of the catalytic converter. It is necessary

to utilize a system approach in prolonging the life of catalytic converter.

Taiwan manufacturers have shown that the catalytic converters can be

durable up to 15,000 km by conducting road tests on 2-stroke motorcycles.

Long-term durability is expected to be better with ceramic catalytic converters

because of superior wash-coat adhesion.

The second aspect of durability is the mechanical integrity of the catalytic

converter. Thermal and vibration conditions that prevail in the exhaust system

of 2-stroke engines can cause cell wall deformation failure of joints at the cell

junctions or separat1on/slid1ng of the honeycomb substrate from the can. In

hot vibrat1on tests (1 050 °C and 75g's) conducted, at the Southwest Research

Institute. d1stort1on of cell walls was observed in metallic substrates. This

XI

deformation of cell walls increased the pressure drop by 30%. Since

temperatures as high as 1050 °C were measured inside the catalytic

converter of the

Kanthal heating Force

element

~-~···at

Thermocouple location Substrate

Figure 3.14: Ceramic converter with heating elements.

110 cc 2-stroke motorcycles, cell deformation can be an issue with metallic

converters. In the case of ceramic converters, cell deformation does not occur

up to 1200 °C.

Separation of the honeycomb substrate from the stainless steel can could

occur due to the failure of brazed joints in the case of metallic substrates or

due to mat failure in the case of the ceramic substrate. To assess mat

durability. three different laboratory tests were employed. The converter was

heated by installing electric heating elements in the periphery of the ceramic

substrate (85 x 80 mm ) as shown in figure 3.14 A tourniquet method was

used to mount ceramtc substrates in 409 stainless steel cans. Temperature

was measured in the outermost cell half way into the substrate, figure 3.15 A

heat shield was provided around the converter to raise the can temperature.

In all the tests, substrate/maUcan interfaces were subjected to shear stress by

applying a load to the substrate in the direction of the honeycomb channels. A

linearly increasing load was applied in the first test to measure the maximum

load required to push the substrate out of the can (high temperature shear

test). In the second test, a static load (45.5 kg) was applied for 100 hours

(static fatigue test). The shear stress at the substrate/mat interface due to this

load was 20.8 KPa. In the third test, a cyclic shear stress was

Shear strength

Static fatigue

Load Cyclic fatigue

Time

Figure 3.15: Schematic load-time curve in the three laboratory

tests.

Applied to the substrate/mat interface to simulate the cyclic nature of vibration

loads (cyclic fatrgue test) The three loading patterns used in these tests are

illustrated m figure 3 15

The effect of temperature on the shear strength of the ceramic converter

package, i.e., the maximum shear stress required to push the substrate out of

the can, is shown in figure 3.16. With increasing test temperature, the shear

strength of the converter package decreases. The decrease in the elastic

properties of the· mat at elevated temperatures is the primary reasons for the

observed decrease in shear strength. The maximum expected shear stress is

14Kpa from the data in figure 3.16 it is evident that the measured shear

strengths are higher than the design stress by more than a factor of 2.

1\ 100 : 90

;;:_ 80 -"' 70 J::: 60 c, 50 t: ~ 40 y; 30 :;; 20 ~ 10 rn 0

-. i f-.

' i

"' l ~

···~

' ~ l

"~ i

""· ' ·~ ~ .. ~

'

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~

Substrate Skin Temperature (oC) ·->

Figure 3.16: A Graph of Shear strength Vs. Temperature (0C).

Static fatigue tests were conducted at 900'C and 1 050'C by applying a

constant shear stress of 20.8kPa. This stress is 50% higher than the design

stress. Five specimens were tested at each temperature. To make the test

more severe. the converter was cooled to room temperature every 10 hours.

The test was discont 1nued 1f the substrate slipped or when the total test time

X·l

The test was discontinued if the substrate slipped or when the total test t1me

reached 100 hours. Substrate slippage did not occur at this stress level up to

1 OOhours at both the temperatures. To assess mat degradation, shear

strength of the ·fatigued specimens was measured at 550'C under isothermal

conditions using a furnace. Figure 3. 17 shows shear strength of the converter

package before and after the static fatigue test. The degradation in the initial

shear strength of the converter package is less than 8%, indicating that the

mat is durable.

A triangular wave load cycle at a frequency of 1 Hz was used in the cyclic

fatigue tests. At 900'C three specimens were cooled to room temperature at

10 hours intervals. The tests were stopped at 100 hours and as before the

residual shear strength were measured at 550'C. No degradation in the shear

strength of the converter package was observed after the 900'C cyclic fatigue

tests. Tests were also conducted at 1050'C under a stress cycle of -15.6 to-

15.6 KPa and the specimens survived for more than 1.3 million cycles (>360

hours) without slipping from the can.

Total time= 100 h

I) _:_ 40 "' ~ 35

~ .30

~ 25 Ol)

ro 20

£ 15 Ol

~ 10 ~

ii) 5 ~

~ 0 .<: (f)

" " " ' " " " , . ,,

. <:> .._<:><:> <f'<:> .,~:~<:> ~<:> .,~:~~:~ ro<:><:> '\~<:> co<:>~ o,<:><:> <:><:><:>

" Substrate Skin Temperature (oC) -->

Temperature cycle = 1 0 h on /1 h off.

X'i

I .

" I

Figure 3.17: Shear strength of ceramic converter package

before and after static fatigue test.

The above data suggest that the mechanical integrity of the ceramic catalytic

converter is adequate to withstand the thermal and mechanical stresses

present in the exhaust systems of 2-stroke motorcycles and scooters.

Another issue that needs to be discussed is the rise in temperature of the

exhaust gases due to exothermic heat generated in the converter. Depending

on the speed of the veh1cle. the rise in exhaust gas temperatures can be 250-

400'C. The temperatures on the surface of the exhaust pipe can reach 700-

sso·c especially 1n w1de-open throttle condition For safety reasons these high

temperatures need to be Isolated from the passengers and the vehicle

footboard Additionally tt1e m1ld steel that 1s used in most exhaust systems

will not last long at these temperatures. It is therefore, necessary to use

stainless steel exhaust pipes with heat shields to address durability and safety

concerns.

3.12 Turbulent Catalysts

Mass transfer for regular catalyst shows a laminar flow after intrusion in the

catalytic converter after a short time. If the change in the metallic substrate

allows the change from laminar flow into turbulent one, longer residence time

and more unconverted gases from the core of the channel come closer to the

catalyst surfaces and more reaction takes place. The higher flow velocity

corresponds to a higher mass transfer. The correspondence between flow

velocity and mass transfer is shown in Figure 3.18.

r1~" ,.,;;n, lt•·'l.~., ...

Figure 3.18: Flow velocity- Mass transfer diagram

The scope of work m this publication covers the analysis of turbulent catalyst

with transversal foil structure (TS) and longitudinal foil structure (LS) with the

focus for use in 2 & 3 wheeler vehicles

X7

3.13 TS- Structure

Transversal Foil Structure (TS) catalyst is the second generation of metallic

substrates. The first generation had straight and unstructured channels. This

TS design is · commercially produced and widely used in regular mass

production for automobile application. In this type, the corrugated foils are

embossed with secondary micro- corrugations (Figure 3.19), which are

provided transverse to the direction of flow i.e. at 90 degree to the flow

direction (see above described mass transfer to the wall).

Figure 3.19: TS structure design with flow details

These micro-corrugations help intense exchanges of unconverted gases in

the core of channel with the converted gases close to the walls.

3.13.1 LS- Structure

Longitudinal Structure (LS) is the third generation of metallic substrate

structure. which has been developed by EMITEC. In the LS design the

corrugated foil is characterized by additional cuts and depressions to provide

shovel like shapes (Figure 3.20)

Figure 3.20: LS structure design with flow details

These counter corrugations projecting into the basic channels create the

effect of additional channels within the same given volume, which results in

turbulent mass transfer to the channel walls and an increased catalytic

reaction . Thus, without increasing the actual surface area of a catalyst, a

higher catalytic efficiency is achieved by the LS design.

3.13.2 Test Bench Analysis

During stationary and dynamic tests catalysts with applied TS as well as LS

metallic substrate structure have been analyzed. To exclude the influence of

production, all converters were preconditioned 4 hours under 950°C ambient

air.

Converters with dimensions as defined in table beneath were analyzed on test

bench:

~

c:~ J::. f,-.

0 a: ~ 0 al 0 £ > _ _, ... (l_

II) )( ;;; :1 Q) ~ U> c: '- a. u f g; c: .... I) 2 ::;) ~ ~ ii E - iii <.) nl "" Of > 0 Q

cu (,) _, §,

33x60 100 Std 5 1 . 4 1 25

33x60 100 TS 51 . 4 1 25

3311:50 8 100 TS 43 -15 •(; 4 I 25

33x508 100 LS 43 - 15% 4 1 25

Table 3.6: Dimensions of tested catalytic converters

All substrates had the same diameter, cpsi, coating and loading. For

analyzing the influence of the structure in combination with the substrate

length, two shorter metallic substrates were applied. Besides, the benefit of

15% reduction· in substrate volume could be achieved. All catalysts were

conditioned at 950°C, 4h in ambient air.

3.13.3 Stationary Measurements

The main task of the engine test bench analysis was to define the exact

conversion rates for several lambda values and different engine speed and

load. Lambda is defined as the actual air-fuel ratio divided by the

stoichiometric air-fuel ratio. If e is smaller than 1, the engine runs with air

deficiency or rich mixture, if e is more than 1, the engine is running lean. The

definition of lambda is shown in the equation below:

[ J -AFR AFR st =e Equation 1: Lambda e................ . . . . ........... lambda[-] AFR ........... air-fuel ratio in the cylinder[-] st AFR ......... stoichiometric air-fuel ratio[-]

Thus a 50cc 4-stroke engine with a free applicable ECU with the possibility to

focus on the selected operation points was mounted on the test bench (Figure

3.21 ).

Figure 3.21: Engine with special exhaust system mounted on engine test bench

The engine configuration is listed in Table 3.7.

Prototype 1 cyhnder 4 stroke eng1ne

Displacement {cm3] 49.8

Bore x Stroke [mm) 40 X 39.8

Compression ratio f ·) 11.1

Coohng system 1quid

v'alve tra.n ~ Valves , OHC. chair

Carburet1on [:-uel inJe~t1on

Table 3. 7 Engine operation points driven on engine test bench

90

In addition, a special assembly of the exhaust system to meet all the

measurement conditions was designed. An important requirement was the

possibility to measure the exhaust gas components before and after the

catalytic converter at the same time during engine operation. Thus, special

heated exhaust pipes, for simultaneous measurement before and after the

catalytic converter, were applied between the exhaust system and the

exhaust gas analyzer. The temperatures before and after catalyst were

examined in the same areas In Table 3. 7 representative engine operation

points, as driven on the engine test bench, can be seen. These points

91

represent the most common operating engine speed and load points for a

scooter vehicle during the ECE-R47 homologation cycle.

Lambda Eng1ne speed [rpm) • BIJIEP fbar)

0.9 5000 rpm - B bar 7000 rpm - 8 bar

1.0 5000 rpm - 8 bar 7000 rom - 7 bar

1.1 5000 rpm - 8 bar 7000 !pm - 5 bar

Table 3.8: Map points for stationary measurements Lambda 0.9

The results of rich operation points represent acceleration phase of 2 & 3-

wheeled vehicles. Turbulent catalysts with different substrate lengths were

compared with the standard catalyst between 5000 and 7000-rpm engine

speed. The standard catalyst shows moderate conversion rates for all three-

emission factors at 5000-rpm operation point (Figure 3.22). Turbulence in the

substrate improves the conversion efficiency for more than 100% in the case

of HC and CO emission factors. Even better conversion rates are represented

for NOx, where a conversion rate of about 50% can be reached. The best

results are presented for the 0 33 x 60 mm catalyst with

TS substrate.

Figure 3.22: Conversion rates 5000 rpm I 8bar I Lambda 0.9

92

Figure 6 shows the results of the 7000-rpm rich operation point. The best HC

conversion is achieved by TS 0 33 x 60 mm substrate all other turbulent

substrates show lower conversion rates. A conversion improvement for TS as

well as LS substrate is seen for the CO and NOx emission factors.

J ~·

§ro~~~~L---------~ !~.~ -~--------------~~ B~r-------------~~ ~ 04-t------------~ >Jt-------------8 tot--.,.----


In general, all conversion rates for rich operation points could be improved

with the turbulent catalysts. The eye-catching improvement is visible for the

NOx conversion rates.

3.13.4 Lambda 1.0

Lambda 1.0 is a typical operation point for the 20-km/h sections in the ECE

R47 homologation cycle (see Figure3t~). This driving point has about 32%

share of the emission factors in the homologation cycle. Standard catalyst

reaches approximately 30% conversion rates for HC and CO, whereas all

turbulent converters exceed these conversion efficiencies (Figure 3.24).

Compared to lambda 0.9, all catalysts reach much higher conversion rates for

HC and CO. NOx conversion for turbulent substrates reaches remarkable

60%. Conversion Rates: 5000 rpm I 8 bar

93

Lambda=1.0

Preconditioned: 4h I 950°C Ambient Air Ageing


A similar behavior than at the operation point with lambda 1.0 and engine

speed 5000 rpm and at lambda 1.0 I 7000 rpm operation point (Figure 3.25)

can be observed. Higher conversion efficiencies were reached for all three

emission factors.

~

j~~~~~~~---=·

•"'!-------., t. t----=.----=-~:AJ ~ P' ~ " ..


Also at this operation point the TS 0 33 x 60 mm substrate shows the best overall

conversion efficiency, compared to other converters.

3.13.5 Lambda 1.1

Engines mounted in a scooter vehicle usually reach their break away point at

lean engine operation to allow use of oxidation catalysts. Therefore these

results represent typical engine operation at wide-open throttle and break

94

away point. The conversion rates for HC and CO show respectable results.

Compared to the standard catalyst (Figure-3.26), TS 0 33 x 60 mm substrate

reaches 25 % improvement of conversion efficiency. The NOx conversion of

TS and LS substrates stagnates.


In the case of high engine speed and lean operation (Figure 1 0), high CO

conversion is representative for all catalysts. Moderate NOx conversion for

two turbulent 0 33 x 50.8 mm substrates can be seen. TS 0 33 x 60 mm

substrate shows much higher HC conversion rates than the standard one.

:! ~~~:=L--1-=--------j " ,.,~-IIJ----

" -~------=--t~+--~ i"

10


It can be stated, that the turbulence in the channels (TS substrate) improves

the overall efficiency of all operating points about 33% at the same substrate

length compared with standard substrate. Shorter TS substrate (0 33 x 50.8

mm) shows similar or slightly better conversion rates than the standard one.

95

LS substrate with 0 33 x 50 8 mm d1mens1ons enhances the convers1on

efficiency compared to the same-dimensioned TS substrate.

3.13.6 Turbulent catalyst backpressure

Exhaust gas mass flow of the small capacity engines rarely reaches values

over 20 kg/h. FigureJ~9 shows, that the backpressure for these values is the

same for turbulent as well as for standard substrates. The backpressure for

significantly increased mass flow values shows a disadvantageous

backpressure behavior of turbulent substrates. No increase in fuel

consumption would be observed for turbulent structures.

,_ , ~ ..; ti • ·•: '" .:br>: ,;ul J

-~ '\; : ~A:;.,, r~ I _/ . 1 ' ·II ( "\

v / 1 .---/_ I ~

/ .L---1 ~ . ' > •

:-""' Ui) ..:4 ~ .... , ... ..I

~\oi!IV.Y ~., .. ~

Figure 3.28: Catalyst backpressure as function of the mass flow

3.13. 7 Dynamic measurements

The modified exhaust system, already used for stationary measurement, has

been ~pplied in a series 50cc 4-stroke scooter vehicle (Figure 3.29). The

technical data of the tested vehicle are presented in Table 3.9. The dynamic

measurements were carried out on a chassis dyno test bench with a Constant

Volume Sampling (CVS) device to detect the exhaust emission

characteristics.

Figure 3.29: Measurement setup for dynamic Tests

Series 50 cc vehJcle V•'ltl, 1 cyl1nder 4 stroke ~ncune (Eum 2 llomolooauon)

r=:arourellon system Constant pressure ca1 bttretor

::oolinq system Air !=>dl aust gas after Ox1dalion catalyst and reatment secondarv ah ~o'd start syslem Tf'lermo s,.~ltch ct ott:e

lr tans1russion CVT secondary gear •JehJcle reference

170 [kg] 1l8SS

Pov.·er ot.~lput 2. 54 [k!~\1 at 7250 [rp1'lj

1.4ea:l fuel con-surnptioo 111 the 42.7 [ktl/1) ht:Pnoloqabon cvcle Homologation Euro 2

.. . . . . . ..

Table 3.9: Technical data of the test vehicle

The standard ECE R47 driving cycle was used for the testing program, in

which the same catalysts were tested during the stationary measurements.

The complete homologation cycle ECE R4 7 consists of eight identical cycles.

The first four cycles are warm up cycles for the homologation

procedure and are at the same time used for sampling the

1)7

exhaust emissions for the hypothetrcal cold start phase for Euro 3. The

complete ECE R4 7 homologation cycle rs shown in Figure 3 30

r: .... , , ........... ••a ·~·::~:..·,-..:.~.~~ .. . .. - ........... -.-· ,...., ,..., ,..., ,...., ,...., ,..... ,...,

! .. I ~

! -~

t ~· 1 l l ...., ....,

l 1 '1 .. . • ,,., , ... .;., • ,., ..

...... f .. c:

Figure 3.30: ECE R47 homologation cycle with defined cold start phase

All emissions, measured separately for the warm and the cold start cycle, were

collected in the emission bags as well as recorded with an online measurement tool.

The emission factors (HC, CO and NOx), resulting of the bag measurements, were

analyzed for each catalytic converter separately. The online measurements provided

an insight into cold start behavior and allowed detailed analysis of the warm-up

phase for tested catalysts.

3.14 Measurement Result

The standard catalyst serves as basis for the comparison with turbulent converter. In

the cold start cycle slightly improved CO emission factors of the TS and LS substrate

(0 33 x 50.8 mm) can be seen. TS substrate with the dimensions 0 33 x 60 mm

shows the best CO emission factor of all substrates. HC+NOx are for both catalysts,

LS and TS with dimensions 0 33 x 50.8 mm, about 18% lower compared to the

standard catalyst. TS substrate with dimensions 0 33 x 60 mm shows 38% better

HC+NOx values than the standard catalyst.

'JX

Figure 3.31 : ECE R47 cold start cycle results

The emission factors in the homologation cycle (Figure 3.32) are limited according to

the Euro2 legislation to 1.0 g/km CO and 1.2 g/km HC+NOx. The focus of the

analysis was not put on the achievement of homologation limits, but to clearly point

out the difference in the catalyst performance. Due to the fact, that standard catalyst

exceeds the limited CO value, a standard catalyst without coating has been used to

make the analysis of turbulent catalyst possible. Nevertheless the standard catalyst

with coating reduces the CO emissions to the half and lowers the HC+NOx. The TS

and LS substrates with the dimensions 0 33 x 50.8 mm show moderately higher CO

values for both short type turbulent catalysts. As expected, the TS 0 33 x 60 mm

substrate shows a 48% better CO emission factor than the standard substrate and is

the only one that falls below the limited Euro 2 value.

Figure 3.32: ECE R47 homologation cycle results

1}<)

The HC + NOx factor of the 0 33 x 50 8 mm TS 1s 14% lower and of the 0 33 x 60

mm LS 32% lower in comparison with the standard catalyst TS substrate w1th the

dimensions 0 33 x 60 mm shows 21% improvement m HC + NOx conversion

3.14.1 Online recorder measurement

The online recorder measurement of lambda as function of the time is shown in

Figure 3.33. With this tool the exact duration of the cold start enrichment phase and

the calibration of the carburetion with secondary air induction were observed.

Furthermore, the predefined lambda values for stationary measurements could be

defined according to the dynamic measurements.

c=J'

'---~ --..- ----.-J . ~ ~ ~ • U

:...tf'-"

Figure 3.33: Lambda history in the ECE R47 cold start cycle

With the help of online measurement temperatures progression before and after

catalytic converter was recorded too. Delta T, presented as substitution between the

temperature after and before the catalyst, shows the starting time of conversion in

the catalyst, which directly impacts the cold start emission factors in the cycle. The

results show that the turbulent LS substrate starts the conversion after 150 s in the I

second cycle followed by the TS (0 33 x 50.8 mm) substrate. The standard

converter starts with conversion after 400 s in the fourth cycle. The eye-catching

100

effect is the peak of the delta T for LS substrate that is twice as h1gh as for the

standard converter, which can be traced back to higher turbulence in the substrate

channel and subsequently a higher mass flow to the substrate walls. The TS 0 33 x

60 mm substrate shows the highest peaks of delta T in the cycle. It starts with the

conversion at the same time as LS substrate.

Figure 3.34: Delta T flow during the ECE R47 homologation cycle

The fuel consumption in the ECE R47 homologation cycle can be seen in Figure

3.35. Due to the fact that similar fuel consumption values can be seen for all

converters, no influence of the turbulent substrates on the engine operation

compared to the standard catalyst was detected. This can be attributed to the

backpressure results, which are shown in Figure 3.28, where no significant

backpressure change at maximum engine mass flow (20 kg/h) was measured.

::

•. r . "

Figure 3.35: Fuel consumption in the ECE R47 homologation cycle

design sr analysis catalytic converter -...

Documents