V8 Engine Breathing Revisited - Gamma Technologies

Autor - Dateiname.ppt 1

GT-SUITE Conference 2007

Frankfurt/Main 2007-10-08, Birmingham/MI, 2007-11-13

V8 Engine Breathing Revisited A GT-POWER Analysis of AFR Control and Performance Issues

Christof Schernus

FEV Motorentechnik GmbH, Aachen, Germany

[email protected]

� Acknowledgements:

� Mr. Andrea Dutto, FEV Motorentechnik GmbH

� Prof. Federico Millo, Politecnico di Torino

This presentation is intended to be a compilation of V8 engine related issues that

may be known to most engine developers who have already dealt with this type of

engines. Nonetheless, talking to professionals around in the industry, I found, V8

engines are often overestimated regarding the uniformity of their cylinder process.

Those deviations from the engine average of individual cylinders will be the focus

of my presentation.


FEV-V8-Breathing.ppt / Schernus / 20071008

2

V8 Engine Breathing Revisited


© by FEV – all rights reserved. Confidential – no passing on to third parties

Contents

Introduction

Basic Exhaust Manifold Layouts

V8 Engine Firing Orders

Full Load Engine Breathing

Air Flow Modeling for Controls

Exhaust Oxygen Sensing

Summary

Here is a brief outline:

After the introduction, I will mention some basics about exhaust manifold layouts.

Then, different firing orders of V8 engines will be addressed.

They will cause differences in engine breathing behavior at full load and part load

that is of concern for air flow control.

Further observations refer to oxygen sensing.

A summary closes the presentation.



3




Introduction

First V8 engines came up 102 years

ago

� 1905: Rolls-Royce V-8

� 3535 cm³, 120° flat-plane

� Intended to replace electric cars

� 1914: Cadillac Type 51 with L-Head Engine

� 341 in³ (5588 cm³), 90° side-valve, 70 hp

� First cross-plane V8

� First mass produced V8

8-cylinder engines are appreciated as

a prime mover of superior comfort

� 90° firing interval ⇒ Small torque fluctuations

� Cross-plane enables second order mass

balance w/o balance shafts

� Sound

Rolls-Royce V8 1905 [1]

Cadillac Type 53 Cabriolet, 1916 [2]

Introduction






Summary

V8 engines are more than 100 years old. The first one was built by RR in 1905. It

was a 3.5 liter 120° V engine and like all early V8 engines it had a flat crankshaft

just like a double straight-4 engine. Interestingly, RR wanted to replace electric cars

that were much common that time. So to be competitive, the engine had to be very

smooth, silent and smoke-free.

The first cross-plane crankshafts came up in 1914 with the famous Cadillac L-Head

engine sharing a patent with Peerless. It was the first mass produced V8 engine, and

in the next decades America became the home of V8s.

So since its first introduction, 8-cylinder engines were appreciated as a prime mover

of comfortable and luxury and/or sporty cars.

The short firing interval produces small torque fluctuations, the cross-plane

crankshaft design enables balancing of 2nd order forces and many love the sound of

V8s, too.



4




Introduction

Despite of their smooth torque

delivery, V8‘s may sometimes be

cross-grained

� Flat-plane crankshafts need balance shafts for

2nd order mass balance like straight-4’s

� Cross-plane crankshafts have large inertia

(counterweights against 1st order moment)

and odd firing along each bank

Odd firing of cross-plane crankshafts

may reportedly cause non-uniform air

supply

� Although as old as cross-plane V8s, this effect

shall be revisited with respect to engine

control

WOT Volumetric Efficiency

Source: Millo et al

GT-SUITE User‘s

Conference, 2003

WOT Volumetric Efficiency

Source: Millo et al

GT-SUITE User‘s

Conference, 2003

Introduction






Summary

[3]

Nonetheless, V8s may surprise us with some characteristics that do not fit very well

with these expectations.

Flat-plane crankshafts without balance shafts have vibrations just like straight-4

engines.

Cross-plane crankshafts need large counterweights against 1st order moments

causing a larger inertia, and they have odd firing intervals along each bank.

This can cause significant differences in volumetric efficiency, as was shown four

years ago by Prof. Millo in his investigation of the Maserati Spyder engine.



5





General Rule for Firing Order Exhaust Manifold:

� Join primary runners of cylinders without interference of exhaust events

� Most preferred, if 2 x 360° or 3 x 240°

� Periodical blow-downs eliminate 0.5th order and provide preconditions for equal gas exchange for each cylinder

� Typical arrangement for V8 with cross-plane crank shaft: 270° + 450°

� Base harmonic = 0.5th order

� Different exhaust pressure level for each

cylinder during exhaust and overlap

� (Characteristic V8 sound)

1 2 3 4

1 2 3

4 5 6

5 6 7 8

1 2 3 4Introduction






Summary

There is a simple rule for exhaust manifold layout:

When connecting primary runners, try to avoid interference of exhaust events.

Three or six cylinder engines with 240° intervals make this pretty easy. Straight-4’s

allow to join cylinders with 360°. All of these solutions allow to eliminate the half

order pressure fluctuation, because the firings of connected cylinders are periodical

and equidistant.

Conventional manifolds for cross-plane V8s allow to connect pairs of cylinders with

270 and 450°, respectively. The unequal intervals does not enable the extinction of

the half order. Therefore, the exhaust pressure level during valve overlap is different

for each cylinder.



6




5 6 7 8

1 2 3 4


„Natural“ V-Angle: 90°

� 90° firing interval without split-pin crank throws

Cross Plane Crankshaft

� Complete 2nd order mass balance

� Bank-wise odd firing featuring intervals

of 90°, 270° and 180°

Flat Crankshaft

� Basically two I4 engines

� Equal firing intervals @ each bank

� Balance shaft required for

2nd order mass balance

1,5

4,8

2,6

3,7

1,5

4,8

2,6

3,7

1-8-6-2-7-3-4-51-8-4-5-7-3-6-21-5-4-3-7-2-6-8

1 2 3 4

5 6 7 8

1,54,8

2,63,7

1-6-3-5-4-7-2-81-6-3-8-4-7-2-51-7-3-5-4-6-2-81-6-2-8-4-7-3-5

RLLRLRRLRLLRLRRLRLLRLRRLRLLRLRRLRLLRLRRLRLLRLRRL

RLRLRLRLRLRLRLRLRLRLRLRLRLRLRLRL

Introduction






Summary

The „natural“ V-angle of an 8-cylinder engine is 90°. It allows to split the 720°

degrees of a cycle in steps of 90° without need for split-pin crank throws.

The mass balance of a cross-plane crankshaft is paid for by odd firing intervals of

90, 180 and 270° on each bank.

(*) The many possible firing orders depend on the arrangement of cranks and sense

of rotation. But their basic pattern remains the same. After alternating the banks, a

double firing occurs on the same bank, as you can see from the black letters in the

Left-Right sequence.

A flat-plane crankshaft makes each bank a inline 4-cylinder engine with equal firing

orders. The banks alternate their firing. First order mass balance comes for free, but

second order forces require balance shafts for comfort.

(*) denotes a keystroke or mouse-click to initiate an animation



7





Flat-plane Intake and Exhaust Flow Rates

1-6-3-5-4-7-2-8

(counterclockwise)

Each bank:

� ∆ϕ=180°

� 1-3-4-2-1

6-5-7-8-6

Introduction






Summary

In a GT-POWER model of a flat-plane crankshaft engine, the valve mass flow rates

of the different cylinders highlight the even firing order of both banks.

Basically, both banks here have a 1-3-4-2 firing order, just in this case the left bank

is delayed by 270°.



8





Cross-plane Intake and Exhaust Flow Rates

1-8-4-5-7-3-6-2

(counterclockwise)

Right bank:

� Cyl. /∆ϕ=

� 1-4: 180°

� 4-3: 270°

� 3-2: 180°

� 2-1: 90°

Left bank:

� Cyl. /∆ϕ=

� 5-7: 90°

� 7-6: 180°

� 6-8: 270°

� 8-5: 180°

Introduction






Summary

In the engine model with cross-plane crankshaft, you find the short firing intervals

between cylinders 2 and 1 on the right-hand side, and between cylinders 5 and 7 on

the left-hand side.

A long break occurs between the blow-downs of cylinders 4 and 3 and between 6

and 8.



9





Flat-plane vs. Cross-plane exhaust pressure

Flat-plane

� Rather uniform

pressure pulses

� Constant average,

no 0.5th order

Cross-plane

� Backpressure drops

btw 4-3 and 6-8

� Short interval btw 2-1,

5-7 leads to higher

backpressure for

cylinders 1 and 7

� FFT results in

remaining 0.5th order

4

3

21

8

6 57

Introduction






Summary

1 3 4

2

65 78

Looking at the exhaust pressure in front of the catalyst, you find two times four (*)

rather equidistant pressure pulses of similar magnitude for the flat-plane engine,

plotted in red. The low pass filter reveals no half order, but a (*) constant average

pressure.

The cross-plane engine has that gap after the pulses of cylinders 4 (*) and 6, in

which the manifold is depleted, and the pressure falls. But during the shortening

firing intervals after that gap, the manifold is recharged with exhaust gas and hence

the (*) pressure increases again to its maximum at cylinders 1 and 7, i.e. those

cylinders blowing 90° after their predecessors. (*) This low frequency pressure

fluctuation from odd firing is noted as half order.



10





Flat-plane WOT Volumetric Efficiency Scatter

Flat crankshaft has

some scatter

� Compact exhaust

manifold with different

primary runner lengths

� Side mounted throttle on

intake plenum causes

different resonances

along entire intake

system

Introduction






Summary

Because of its more uniform backpressure, the engine with flat-plane crankshaft has

just a small scatter of full load volumetric efficiency. Intake cam phaser angle was

optimized at each speed in a trade off between volumetric efficiency and cam phaser

excursion. The observed variation of volumetric efficiency is related to intake and

exhaust manifold asymmetries. (The compact exhaust manifold features different

primary runner lengths. And feeding the intake plenum through a side mounted

throttle creates different system modes or wavelengths, respectively, between

cylinders and upstream pressure nodes, like air cleaner or snorkel inlet.)



11





Cross-plane WOT Volumetric Efficiency Scatter

Cross-plane has

significantly higher

scatter

� same intake manifold

� 4-in-2-in-1 exhaust

manifolds combining

cylinders with 270/450°

intervals

5 6 7 8

1 2 3 4

Introduction






Summary

The intake system is the same for the cross-plane engine. And the exhaust manifold

combines primary runners of large firing intervals. Nonetheless, we see a large

scatter band of volumetric efficiency. The right bank shows a (*) significant non-

uniformity at 3500 rpm, that we will take a closer look at.



12





Cross-plane gas exchange at 3500 rpm, WOT, Cylinder 1

Introduction






Summary

The valve timing is set equal for all eight cylinders.

Looking at the intake and exhaust pressures at cylinder 1, we see a (*) suction wave

coming back to the exhaust valve while the intake pressure is on a higher level. The

gas exchange profits from this pressure gradient by (*) scavenging residuals as

indicated by the positive mass flow rates in intake and exhaust.



13






Introduction






Summary

The situation is the same for cylinder three



14






Introduction






Summary

And can still be found at cylinder 4, although a bit weaker



15






Introduction






Summary

But it is just the contrary at cylinder 2. Because the exhaust valve of cylinder 1

opens just 90° after cylinder 2, its blow-down pressure pulse hits cylinder 2 in the

(*) last 45° of its exhaust stroke. It causes even some (*) backflow from the exhaust

into the cylinder against the upward moving piston and spoils the overlap. Instead of

residuals scavenging, we see backflow of exhaust gas into the intake system. And

where exhaust gas has to be ingested during suction, there is no more room for fresh

air. This explains the low volumetric efficiency of cylinder 2.



16





Comparison of cross-plane and flat-plane engine

� Cross-plane exhibits

significantly larger

standard deviation

at certain engine

speeds

� Issues:

� Knock control

� AFR uniformity

� Misfire detection

� Comfort

Introduction






Summary

All over the speed range of our example engine, we see no clear drawbacks or

advantages for either firing order regarding average volumetric efficiency and

residual gas fraction. But there is a clear disadvantage for the cross-plane engine in

terms of standard deviation. While the flat-plane engine stays below 2%, the cross-

plane always exhibits much higher or in the best case equal standard deviation of

volumetric efficiency. Similar deviations apply to residuals. From this non-

uniformity follow differences in AFR, knock behavior, and – under certain

circumstances – fluctuations of torsional speed and vibrations.



17





Approach based on filling straight line

p0pV

∆∆η

Two stage model: ( )( )

( )( )VVTrpm

VVTrpm

VVVTrpmp pp

pEffVol ,0

,

,,. −

∆∆

=η

Cross-plane2000 rpm

Introduction






Summary

Now this may pose challenges to engine air flow control, too. Engine controls often employ two

stage models for air flow.

To create a two stage model, volumetric efficiency is measured in manifold pressure sweeps at

constant valve timings and fixed engine speeds. Typically a linear correlation of volumetric

efficiency and manifold pressure is found. At a still positive pressure p0 (*), the extrapolated

volumetric efficiency would be zero, and the slope of volumetric efficiency should be a constant

gradient (*).

These two parameters of the linear equation, p0 and gradient, have to be found for any engine

speed, intake valve lift and intake and exhaust cam phasing.

Please note that for small valve overlap, we see just a small scatter of these characteristics for

the different cylinders even in case of the cross-plane crankshaft.



18




Local model of breathing straight line


Two stage model based on breathing straight line

Manifold

air pressure

Engine

Speed

Var. Valve

Actuation

Global model

vol.eff. gradient

p

V

∆∆η

Global model

Pressure Offset

0p

( )p

pp VV ∆

∆−=

ηη 0

Intake

Lift

Cam

Angles

Introduction






Summary

Now this is the structure of such an air flow two stage model. We don’t build a model of air

flow rate or volumetric efficiency for the complete set of operation point parameters including

manifold pressure. Once the local pressure offsets p0 and gradients of volumetric efficiency

have been determined, we can fit global Response Surface Models to these sweep variables

using polynomials or non-linear functions for the entire range of VVA parameters and engine

speeds.

The resulting two stage model can be used in the ECU or in Hardware-in-the-Loop Simulation

to predict the amount of air ingested by the engine based on VVT settings, engine speed and

manifold air pressure.



19





Linear fit less appropriate at large overlap

� Cross-plane, large

valve overlap

� Curvature causes

problems for linear fit

� Higher order

polynomials no good

option either

Two stage model: ( )( )

( )( )VVTrpm

VVTrpm

VVVTrpmp pp

pEffVol ,0

,

,,. −

∆∆

=η

Introduction






Summary

Unfortunately, that linear correlation is not always sufficiently accurate. Such significant

curvature (*) of the breathing line may be found where overlap cross sections are no longer

small compared to the intake suction area. Higher order polynomials may not comply with the

required monotony of the model and therefore provide unphysical results.



20





Linear fit less appropriate at large overlap

� Cross-plane, large

valve overlap

� Curvature causes

problems for linear fit

� Higher order

polynomials no good

option either

� Cross-plane: large

relative deviations

� Workaround:

Hardwired correlation

of overlap and load

map enables linear

approach for air flow

with local validity

� Reduced overlap at

low load improves

scatter width of ηV and AFR Two stage model: ( )

( )( )( )VVTrpm

VVTrpm

VVVTrpmp pp

pEffVol ,0

,

,,. −

∆∆

=η

-60%

-40%

-20%

0%

20%

40%

60%

80%

100%

0 0.2 0.4 0.6 0.8 1

manifold pressure (bar)

vol.eff. deviation from average

cyl.1

cyl.2

cyl.3

cyl.4

Introduction






Summary

The situation becomes more difficult, (*) because the scatter of the cross-plane V8 engine

produces a relative deviation from average volumetric efficiency that increases towards low

manifold pressures. This may result in large lambda changes and misfire. Similar reasons may

be found for poor idle stability of V8’s with cross-plane crankshaft.

(*) A workaround is available through a hardwired correlation of cam phasing and target load,

significantly reducing overlap toward lower manifold pressures. Then, the linear equation may

still be valid in a limited pressure range around the target value. (*) And the scatter of

volumetric efficiency can be handled, too.

Let me add the note that the flat-plane V8 engine is much less sensitive to this effect due to its

even firing order.



21





TWC has to operate at λλλλ=1� Cylinders deliver different amounts of gas at different λ� Some run rich, others lean

� Catalyst relevant λ is mass average:

EGO sensor signal influenced by oxygen partial pressure @ ZrO2 sond

� Effected by exposure time fraction of exhaust gas

� Under stationary conditions, cycle average λ signal should be proportional to the average cylinder mass fractions in the EGO body

� Transport of gas from different cylinders into the lambda sond plays important role

( )∑

∑=icyl

icylicyl

m

m

,

,, λλ

∫∫∑

∫∫∑ =∝

dt

dt

dt

dt icyl

icyl

icylicyl

EGO

,

,

,, ξλ

λξλ

Introduction






Summary

The AFR excursions lead to the next paragraph, how to obtain reliable information

about at which lambda the engine is operating.

Let me suppose the gases are distributed uniformly in the all cross sections, well

knowing there are many effects related to 3D flow that could blur the 1D findings

we discuss here.

Especially cylinders of a cross-plane V8 may run at unequal lambda. Some run rich,

others lean. If these deviations cannot be reduced, at least the mass average lambda

shall be correct to keep the 3-way catalyst working.

The oxygen sensor however will provide a signal based on the oxygen partial

pressure on the surface of the electrode. And after low pass filtering this signal is

rather a time-mass average of the gases in touch with the electrode.

Now, how do gases get from the cylinders to the electrode?



22





Cylinder Tracking in the exhaust line

Tracer Gas Injection

� Tracer gas is injected into each cylinder before EVO proportional to it‘s amount of fresh air (0.01%)

� This gas can be traced all along the exhaust system

by sensors

� The normalized mass fractions of tracer gas represent

the local composition as of the cylinders

Oxygen Sensor Model

� EGO sensors commonly consist of a ceramic sensor

element inside a metal heat shield

� Exhaust gas is transferred in and out through holes on

the front of the heat shield

� Compression and expansion is the dominant mechanism of mass exchange

Introduction






Summary

To model this transport we inject 0.01% of the ingested air mass as tracer gas into

each cylinder before EVO.

Using sensors, we can trace these gases all along the exhaust system model and tell

at each place and time to which proportion the cylinders have contributed to the gas

mixture.

(Click movie) From CFD we have learned that for the typical oxygen sensors with

holes on the tip, compression and expansion is the dominant mechanism of mass

exchange. Gas will flow into the heat shield and change the composition inside,

when the pressure increases, and the heat shield will be depleted without changing

mass fractions when the pressure falls.



23





UEGO Sensor Model

GT-POWER (U)EGO sensor model

� Volume of heat shield connected to

catalyst inlet cone by orifices

� Lambda sensed inside heat shield

� Tracer gas mass fractions sensed and

normalized

� inside heat shield and

� at inlet cone

� Pressure changes will cause mass

exchange by compression/expansion of

gas inside heat shield

Introduction






Summary

Our GT-POWER model reflects this by a small volume representing the heat shield

connected to the converter inlet cone by orifices – the small holes in the sensor tip.

Here we can now sense lambda and the mass fractions of the tracer gas.



24





Cylinder Signature of Flat-Plane Engine

� 3000 rpm,

BMEP=3 bar

� Flat-plane engine

shows similar

signatures inside

UEGO sensor

heat shield

� Signatures ∝cylinder vol.eff.

Introduction






Summary

At 3000 rpm and 3 bar BMEP we first look at the flat-plane engine. We find rather

similar pressure pulses every 180°.

Every increase of pressure causes flow into the heat shield and will cause a change

of composition.

Because of the uniform transport conditions along the cycle, the cylinder signature

inside the oxygen sensor is closely related to the variation of volumetric efficiency.

These are good preconditions for AFR control.



25





Cylinder Signature of Cross-Plane Engine

� 3000 rpm,

BMEP=3 bar

� Cross-plane

engine: cyl. 4 is

outlier

� More frequent

and larger rises

of pressure when

cylinder 4 gas

present in front

of UEGO sensor

Introduction






Summary

The cross-plane engine does it an other way.

During the rising slope of the half order pressure fluctuation, we see more frequent

and higher pressure increases, and hence, larger mass flow pulses into the sensor.

During this period, cylinder 4 is present in the cone with higher concentration than

the others. Therefore, its concentration in the heat shield is higher, too. The AFR

signal will therefore emphasize the value of cylinder 4 lambda. If this one runs lean,

the AFR control may turn the entire engine too rich. (Which cylinder is over- or

underrepresented depends on the firing order, on the exhaust manifold volume

between valves and oxygen sensor and to some extent also on engine load.)



26





Cylinder Signature Flat Plane Eng. w/ Cross-Plane Manif.

� 3000 rpm,

BMEP=3 bar

� Flat-plane firing

order with cross

plane manifold

⇒ lower deviation

of signatures

� Periodic

pressure

amplitudes w/o

0.5th order do

not emphasize

individual

cylindersIntroduction






Summary

To validate, if this effect would not be caused by improper layout of the exhaust

manifold, we simply ran the flat-plane firing order on the engine with cross-plane

exhaust manifold. And see, the correlation between volumetric efficiency and

cylinder signature in the sensor has become much better.



27




Summary

Cross-plane crankshaft: Odd firing order

� 0.5th order pressure fluctuations in conventional exhaust manifolds

� Different conditions during valve overlap

� Non-uniform internal EGR and volumetric efficiency

� Scatters of volumetric efficiency can be reduced by small overlap

� AFR sensor signal may ignore or emphasize individual cylinders due to coincidence of

pressure pulse frequency and presence of gas in front of EGO sensor

� While differences in air flow may be minimized by proper tuning of manifolds and valve

timing, there is no ultimate panacea for AFR sensor errors

Flat-plane engine

� Preferred solution for sports and racing applications

� Comfort reasons may require balance shaft for 2nd order mass forces

Introduction






Summary

…



28




References

[1] "Rolls-Royce V-8 (1905)." Wikipedia, The Free Encyclopedia. 1 Oct 2007, 09:27 UTC.

Wikimedia Foundation, Inc. 07 Oct 2007 <http://en.wikipedia.org/w/index.php?title=Rolls-

Royce_V-8_%281905%29&oldid=161509935>.

[2] "Cadillac Type 51." Wikipedia, The Free Encyclopedia. 26 Jan 2007, 11:18 UTC. Wikimedia

Foundation, Inc. 07 Oct 2007

<http://en.wikipedia.org/w/index.php?title=Cadillac_Type_51&oldid=103355675>.

[3] „Improving Misfire Detection in an 8-Cylinder FERRARI Engine.“ F. Millo, F. Mallamo, R.

DiGiovanni (Politecnico di Torino) and A. Dominici (Ferrari Auto S.p.A). GT-SUITE

European User‘s Conference, 20 Oct 2003, Frankfurt/Main, Germany. 07 Oct 2007

<http://www.gtisoft.com/confarch/MisfireDetect.ZIP>

Documents

V8 Engine Breathing Revisited - Gamma Technologies