UNITED
STATES
ENVIRONMENTAL
PROTECTION
AGENCY
ANN
ARBOR,
MI
48105
June
23,
2005
OFFICE
OF
AIR
AND
RADIATION
MEMORANDUM
SUBJECT:
Exhaust
Emission
Testing
of
Two
High­
Performance
SD/
I
Marine
Engines
FROM:
Mike
Samulski
and
Matt
Spears
Assessment
and
Standards
Division
TO:
Docket
OAR­
2004­
0008
This
memorandum
describes
exhaust
emission
testing
of
two
"
high­
performance"
marine
engines.
The
two
engines
were
a
carbureted
738
horsepower
and
a
supercharged
and
fuel­
injected
1040
horsepower
sterndrive/
inboard
(
SD/
I)
marine
engine.
The
engines
were
operated
on
a
dynamometer
over
a
five­
mode
steady­
state
marine
duty
cycle.
Gaseous
hydrocarbon
(
HC),
carbon
monoxide
(
CO),
and
oxides
of
nitrogen
(
NOx)
were
measured
using
a
portable
emissions
measurement
sampling
unit
(
PEMS).

At
the
time
of
this
testing,
no
data
was
available
on
emission
rates
from
high­
performance
SD/
I
engines.
The
purpose
of
this
testing
was
to
help
determine
baseline
emissions
for
these
engines.
We
measured
emissions
of
21.6
g/
kW­
hr
HC+
NOx
and
253
g/
kW­
hr
CO
for
the
carbureted
engine
and
12.5
g/
kW­
hr
HC+
NOx
and
349
g/
kW­
hr
CO
for
the
supercharged
and
fuel­
injected
engine.
For
reasons
discussed
later
in
this
memo,
the
measured
HC
emissions
for
the
carbureted
engine
are
probably
understated.

Background
Typical
gasoline­
powered
sterndrive/
inboard
marine
engines
are
marinized
using
automotive
engine
blocks.
Sterndrive
engines
are
those
that
are
mounted
at
the
rear
of
the
vessel
with
a
drive
unit,
attached
directly
to
the
engine,
that
passes
through
the
hull
of
the
vessel.
These
engines
are
also
known
as
"
inboard/
outboard"
engines.
Inboard
engines
are
similar
to
sterndrive
engines
except
that
they
may
be
located
away
from
the
stern
and
are
connected
to
the
propeller
by
a
propeller
shaft
that
passes
through
the
hull.

Marinization
refers
to
making
modifications
and
adding
components
to
the
basic
engine
block
to
create
an
engine
that
is
suitable
for
use
in
marine
applications.
These
modifications
include
choosing
and
optimizing
the
fuel
management
system,
configuring
a
marine
cooling
system,
adding
1
Samulski,
M.,
"
Sensitivity
of
Test
Cycle
and
Fuel
Type
on
a
Spark­
Ignition
Four­
Stroke
Inboard
Marine
Engine,"
SAE
Paper
941782,
1994.

­
2­
intake
and
exhaust
manifolds,
and
adding
associated
accessory
drives
and
units.
In
traditional
SD/
I
engine
designs,
these
engines
are
not
modified
to
produce
significantly
more
power
than
the
base
automotive
engine.

A
small
subset
of
SD/
I
engines
are
designed
to
produce
much
higher
power
than
an
automotive
engine
of
the
same
engine
displacement.
These
engines
are
referred
to
as
highperformance
or
racing
engines.
These
engines
are
often
used
in
racing
applications
which
would
be
exempt
from
exhaust
emissions
under
the
Clean
Air
Act.
However,
these
engines
are
also
commonly
used
for
recreation.
Typical
high­
performance
vessels
are
powered
by
two
or
three
of
these
engines.
Because
of
the
short
time
between
rebuilds
of
these
engines,
it
is
common
practice
for
boat
owners
to
own
to
sets
of
engines
so
that
one
set
can
be
used
while
the
other
is
being
rebuilt.

High­
performance
marine
engines
are
typically
built
up
from
custom
engine
blocks
that
are
the
same
as
or
similar
to
engine
blocks
used
in
race
cars.
These
engine
blocks
typically
have
larger
bores
than
comparable
automotive
engines
and
are
more
robust
so
to
be
able
to
withstand
higher
forces.
Increased
horsepower
is
achieved
in
three
ways:
higher
fuel
rates,
better
air
management,
and
higher
engine
speeds.
These
higher
engine
speeds
require
engine
components
to
be
machined
with
precise
tolerances.
Otherwise,
the
high­
speed
flexing
of
the
parts
can
cause
them
to
fail.
In
addition,
these
engines
typically
run
rich
to
help
protect
the
engine
against
high
temperatures.

Engines
and
Fuel
Specifications
We
tested
two
high­
performance
engines
as
a
part
of
this
effort.
The
first
was
a
carbureted
738
horsepower
engine
manufactured
by
Sterling
Performance.
The
second
was
a
supercharged
and
fuel­
injected
1040
horsepower
engine
manufactured
by
Baker
Engineering.
We
chose
these
engines
to
provide
examples
of
two
widely
different
styles
of
high
performance
engines.
Both
of
these
engines
had
been
operated
in
boats,
but
it
was
not
possible
to
determine
the
hours
of
use
on
these
engines.
Table
1
presents
a
side­
by­
side
comparison
of
the
two
engines
which
are
shown
in
Figures
1
and
2.
Note
that
both
manufacturers
have
diverse
product
lines
and
this
comparison
is
not
intended
to
be
representative
of
the
capabilities
of
each
company.

The
fuel
used
for
this
testing
was
93
octane
Sunoco
gasoline
purchased
at
a
nearby
gas
station
in
Milford,
MI.
Through
a
visit
to
the
gas
station
after
the
testing,
it
was
determined
that
this
fuel
contained
10%
ethanol.
Because
the
ethanol
in
the
fuel
contains
oxygen,
it
would
be
expected
that
the
HC
and
CO
emissions
on
this
fuel
would
be
somewhat
lower
than
on
straight
gasoline.
Previous
testing
on
a
carbureted
5.7
liter
engine
showed
a
10
percent
reduction
in
HC
and
25%
reduction
in
CO
on
E10
compared
to
gasoline.
1
The
oxygen
in
the
fuel
would
also
be
expected
to
increase
NOx
emissions
which
may
offset
the
HC
reduction
in
the
combined
HC+
NOx
measurement.
The
fuel
consumption
calculations
were
made
using
a
specific
gravity
of
0.77
which
was
based
on
the
known
specific
gravity
for
other
Sunoco
fuels.
Sterling
Performance
reported
that
they
observed
a
peak
power
of
1200
hp
on
the
supercharged
engine
when
operated
on
high­
octane
racing
fuel.
­
3­
Figure
2:
Fuel­
Injected,
Supercharged,
and
Intercooled
1040
hp
Engine
Figure
1:
Carbureted
and
Naturally­
Aspirated
738
hp
Engine
Table
1:
Engine
Specifications
Sterling
Performance
Baker
Engineering
Block
Configuration
Displacement
Rated
Power
Rated
Speed
Air
Induction
Oil
System
V8
588
cubic
inches
738
hp
(
550
kW)
5800
rpm
naturally
aspirated
dry
sump
V8
632
cubic
inches
1040
hp
(
778
kW)
5800
rpm
supercharged,
intercooled
dry
sump
Fuel
system
4
Barrel
Holley
Carburetor
throttle
body
injection
(
2
throttle
bodies)
8
fuel
injectors
batch
fire
per
bank
electronic
fuel
pump
Electronic
control
distributer
with
electronic
control
of
ignition
system
Fuel
Air
Spark
Technology
open­
loop
2
Morgan,
E.,
Lincoln,
R.,
"
Duty
Cycle
for
Recreational
Marine
Engines,"
SAE
Paper
901596,
1990.

­
4­
Figure
3:
Sample
Probe
Location
Test
Methodology
Testing
was
performed
at
Sterling
Performance.
Although
Sterling
Performance
is
equipped
to
test
engines
for
power
and
fuel
consumption,
they
are
not
equipped
to
perform
exhaust
emission
measurements.
Exhaust
emission
measurements
were
made
with
a
portable
emission
sampling
system
provided
by
EPA.

Testing
was
performed
over
the
E4
duty­
cycle
adopted
by
the
International
Standards
Organization
to
represent
operation
of
pleasurecraft
with
spark­
ignition
marine
engines.
This
is
the
same
duty
cycle
that
is
currently
used
to
certify
outboard
and
personal
watercraft
marine
engines
to
EPA
standards.
The
E4
duty
cycle
contains
five
test
modes
including
idle,
full
power,
and
three
intermediate
modes
along
a
theoretical
propeller
curve.
2
Table
2
presents
this
duty
cycle.

Table
2:
Modal
Duty
Cycle
for
SI
Marine
Engines
Mode
%
of
Maximum
Test
Speed
(
MES)
%
of
Maximum
Torque
at
MES
%
of
Maximum
Power*
at
MES
Weighting
Factor
1
2
3
4
5
100
80
60
40
idle
100
71.6
46.5
25.0
0
100
57.2
27.9
10.1
0
0.06
0.14
0.15
0.25
0.40
*
%
power
=
(%
speed)
×
(%
torque).

To
create
resistance
(
load)
on
the
engine
during
testing,
the
engine
was
connected
with
a
drive
shaft
to
a
water­
brake
dynamometer.
Fuel
consumption
was
measured
through
the
use
of
a
small
fuel
reservoir
and
two
turbine
flow
meters.
Overflow
fuel
from
the
fuel
injectors
was
returned
to
the
reservoir
to
allow
for
an
accurate
measurement
of
fuel
consumption.
During
engine
operation,
the
test
cell
heated
due
to
engine
operation.
However,
constant
temperature
air
was
drawn
from
outside
the
test
cell
and
routed
to
a
hood
above
the
air
intake
on
the
engines.

In
a
typical
boat
application,
cooling
water
is
drawn
from
the
ambient
water
and
routed
through
the
engine
and
cooling
system.
This
water
is
then
dumped
into
the
exhaust
stream.
In
this
testing,
cooling
water
was
routed
through
the
engine
to
a
heat
exchanger
which
cooled
the
water
and
­
5­
Figure
4:
Oil
Reservoir,
Intake
Air
Hood,
and
Cooling
Fan
returned
it
to
the
engine.
Because
no
water
was
mixed
into
the
exhaust,
a
dry
sample
could
be
taken.
Emission
measurements
were
made
by
inserting
a
sample
probe
into
the
exhaust
manifold
on
the
left
bank
of
the
engine
downstream
of
where
the
exhaust
gases
from
the
four
left
bank
cylinders
mix.
Exhaust
concentration
was
then
based
on
the
sample
from
the
left
bank
and
total
emissions
were
determined
using
the
total
fuel
consumption.
We
first
considered
sampling
downstream
of
where
the
exhaust
pipes
from
the
two
banks
are
joined.
However,
there
were
friction
fittings
(
joining
the
exhaust
manifold
and
corrugated
piping)
upstream
of
this
point
where
air
may
have
been
drawn
into
the
exhaust
stream.
This
air
flow,
which
could
not
be
measured,
would
likely
have
diluted
the
measured
exhaust
concentration
if
we
had
sampled
at
that
point.

Both
of
the
engines
had
external
oil
tanks
with
a
dry
sump
in
the
oil
pan.
A
cooling
fan
was
directed
at
the
oil
reservoir
and
the
engine
during
testing.
As
discussed
above,
the
air
in
the
test
cell
increased
in
temperature
during
testing.
In
addition,
especially
at
high
power
modes,
the
water
temperature
and
oil
temperature
increased
during
testing.
This
increase
in
temperature
was
larger
than
would
be
seen
in
an
in­
use
application
because
of
the
efficiency
of
the
cooling
systems
in
boats
which
have
the
potential
to
draw
large
amounts
of
ambient
water
through
the
engine.
Therefore,
we
used
short
sample
times,
especially
at
the
peak
power
mode,
to
prevent
engine
overheating
At
wide­
open­
throttle,
we
sampled
for
30
seconds.
At
the
remaining
modes
we
sampled
for
two
minutes
each.

Portable
Emissions
Measurement
System
We
used
a
portable
emissions
measurement
system
(
PEMS)
to
measure
engine
emissions.
The
PEMS
we
used
was
a
Semtech­
D
manufactured
by
Sensors,
Inc.,
Saline,
MI.
Semtech­
D.
Table
3
presents
the
emissions
measurement
technology
to
analyze
hydrocarbons,
oxides
of
nitrogen,
carbon
monoxide,
and
carbon
dioxide.
All
of
these
analyzers
are
specifically
allowed
by
EPA
for
conducting
any
emissions
certification
and
compliance
testing.
To
ensure
that
all
of
the
EPArequired
emissions
calculations
were
properly
conducted,
Semtech­
D
also
measured
atmospheric
pressure
and
intake
air
temperature
and
humidity.
­
6­
Table
3:
Emissions
Measurement
Technology
Emission
Analyzer
technology
Comments
Nitric
oxide,
NO
Nondispersive
Ultraviolet,
NDUV
NO
and
NO
2
are
added
to
determine
total
oxides
of
nitrogen,
NOx
Nitrogen
dioxide,
NO
2
Nondispersive
Ultraviolet,
NDUV
Carbon
Monoxide,
CO
Nondispersive
Infrared,
NDIR
Carbon
Dioxide,
CO
2
Nondispersive
Infrared,
NDIR
Total
hydrocarbons,
THC
Heated
Flame
Ionization
Detector,
HFID
HFID
and
all
sample
lines
heated
to
191
°
C
The
Semtech­
D
is
normally
equipped
with
an
averaging
Pitot
tube
exhaust
flow
meter,
which
is
used
along
with
the
emissions
concentration
measurements
to
determine
emissions
rates
in
grams
per
second.
However,
to
facilitate
rapid
installation,
the
flow
meter
was
not
used.
Instead,
Semtech­
D
calculated
grams
of
emissions
per
grams
of
fuel
consumed
based
on
summing
CO,
CO
2
and
THC
concentrations
and
factoring
in
certain
molecular
weights
and
the
hydrogen­
to­
carbon
ratio
of
the
fuel.
To
determine
emissions
flow
rates
in
grams
per
second,
the
fuel­
specific
emissions
data
from
Semtech­
D
were
multiplied
by
fuel
consumption
data
provided
by
a
fuel
flow
meter
that
was
already
installed
at
the
Sterling
Performance
Marine
laboratory.
Brake­
specific
emissions
were
determined
by
dividing
the
emissions
rates
by
the
power
developed
by
the
engine.

The
emissions
test
procedure
was
conducted
as
follows:

1.
Semtech­
D
was
started,
warmed­
up,
and
the
automated
leak­
check
was
performed.
2.
Each
gas
analyzer
within
the
Semtech­
D
was
zeroed
and
spanned,
and
then
audited
with
mid­
span
calibration
gases
both
prior
to
testing
and
after
testing
to
ensure
that
EPA
certification
and
compliance
thresholds
for
analyzer
drift
(
i.
e.,
±
2%
of
span)
were
met
for
each
test.
3.
Emissions
were
stabilized
at
each
steady­
state
mode
and
a
fixed
amount
of
time
was
determined
over
which
fuel­
specific
emissions
were
averaged.
A
shorter
amount
of
time
was
used
during
the
full
power
mode
to
prevent
engine
damage.
4.
Each
test
was
validated
by
verifying
acceptable
emissions
analyzer
drift
and
by
verifying
that
Semtech­
D
did
not
report
any
errors
such
as
any
pressures
or
temperatures
being
out
of
EPA­
required
limits.
3
"
Draft
Regulatory
Support
Document:
Control
of
Emissions
from
Spark­
Ignition
Marine
Vessels
and
Highway
Motorcycles,"
U.
S.
EPA,
July
2002.

­
7­
Test
Results
and
Discussion
Carbureted
Engine
Table
4
presents
the
modal
and
weighted
test
results
for
the
carbureted
engine.
The
combined
HC+
NOx
value
was
measured
to
be
21.6
g/
kW­
hr
and
CO
was
measured
to
be
253
g/
kWhr
These
emissions
are
high
compared
to
traditional
SD/
I
engines
which
have
typically
have
exhaust
emissions
on
the
order
of
14­
18
g/
kW­
hr
HC+
NOx
and
100­
200
g/
kW­
hr
CO.
3
These
higher
emissions
are
likely
due
to
the
high
fueling
rate
combined
with
rich
air­
fuel
ratios
and
low
residence
times
in
the
cylinder.

Table
4:
Exhaust
Emission
Test
Results
for
738
hp
Naturally­
Aspirated
and
Carbureted
Engine
Speed,
rpm
Power,
kW
BSFC,
kg/
hr
HC,
g/
hr
NOx,
g/
hr
CO,
kg/
hr
5802
4639
3481
2321
1345
550
316
153
60
1
164
99
59
30
9
8409
1593
1067
1195
838*
9610
2250
249
60
59
67
58
58
27
4
weighted
result,
g/
kW­
hr
376
13.2
8.4
253
*
may
be
understated
because
exhaust
concentration
was
above
measurement
range
To
protect
against
high
temperatures,
high
performance
engines
are
typically
calibrated
to
operate
very
rich
of
stoichiometric.
This
reduces
the
combustion
efficiency
and
lowers
cylinder
temperatures.
In
addition,
liquid
fuel
helps
absorb
heat
and
lubricate
the
cylinders.
Especially
at
idle,
carbureted
and
throttle­
body
injected
engines
are
often
calibrated
rich
to
prevent
the
risk
of
misfire
due
to
cylinder­
to­
cylinder
variation
in
the
air
fuel
ratio.
This
variation
is
due
differences
to
the
geometry
of
the
air
intake
path
for
each
cylinder
and
to
the
offsets
in
cylinder
valve
and
spark
timing.
For
naturally­
aspirated
engines,
the
air­
fuel
ratios
for
each
cylinder
are
also
sensitive
to
the
pressure
pulses
created
by
the
valve
timing.

At
idle,
the
hydrocarbon
concentration
in
the
exhaust
was
higher
than
what
could
be
measured
accurately
using
the
PEMS
unit.
Therefore,
the
actual
HC
emission
may
be
significantly
higher
than
what
is
reported
in
Table
3.
An
investigation
of
the
air
fuel
ratio
at
idle
suggested
that
raw
fuel
may
have
been
passing
through
one
of
the
cylinders
unburned.
As
a
sensitivity
analysis,
we
consider
the
scenario
where
the
air­
fuel
ratio
in
the
exhaust
at
idle
was
the
same
as
for
the
fuelinjected
engine.
As
shown
in
the
attachments,
the
exhaust
air­
fuel
ratio
for
these
two
engines
is
similar
for
the
other
test
modes.
Under
this
scenario,
the
actual
hydrocarbon
concentration
at
mode
5
would
have
been
50%
higher
than
what
was
measured.
Under
this
sensitivity
analysis,
the
weighted
hydrocarbon
emissions
were
calculated
to
be
14.7
g/
kW­
hr.
­
8­
We
performed
a
second
sensitivity
analysis
in
which
we
looked
what
would
happen
if
the
idle
mode
were
calibrated
for
low
HC.
To
do
this,
we
modeled
the
idle
mode
emissions
based
on
those
for
the
fuel­
injected
engine
(
adjusted
for
engine
speed
and
displacement).
Using
a
value
of
282
g/
hr
at
idle,
we
get
a
weighted
HC
estimate
of
11.3
g/
kW­
hr.
Although
the
idle
mode
is
the
heaviest
weighted
mode
in
the
duty
cycle,
the
actual
mass
flow
is
relatively
small.
Therefore,
even
if
this
change
were
made
at
idle,
the
weighted
HC
emissions
would
still
be
higher
than
typical
SD/
I
engines.

The
high
NOx
emissions
at
idle
suggested
that
there
is
likely
high
cylinder­
to­
cylinder
variation.
These
engines
are
typically
calibrated
for
peak
power.
Air
intake
paths
for
the
carbureted
engine
are
designed
to
use
pressure
pulses
to
help
ram
the
air
into
the
cylinder
at
wide
open
throttle.
Due
to
the
cylinder­
to­
cylinder
variation,
one
cylinder
may
be
running
much
leaner
than
the
others.
It
is
for
this
reason
that
the
engine
is
calibrated
for
such
rich
overall
operation
at
idle.
In
addition,
this
engine
is
designed
with
a
high­
compression
ratio
to
increase
power.
High
compression
ratios
also
increase
peak
cylinder
temperatures
and
NOx
is
formed
at
these
high
temperatures.

Fuel­
Injected
Engine
Table
5
presents
the
modal
and
weighted
test
results
for
the
fuel­
injected
and
supercharged
engine.
The
combined
HC+
NOx
value
was
measured
to
be
12.5
g/
kW­
hr
and
CO
was
measured
to
be
349
g/
kW­
hr.
This
HC+
NOx
level
is
lower
than
most
typical
SD/
I
engines
on
a
brake­
specific
basis.
However,
CO
emissions
are
about
twice
as
high.

Table
5:
Exhaust
Emission
Test
Results
for
1040
hp
Fuel­
Injected,
Supercharged,
and
Intercooled
Engine
Speed,
rpm
Power,
kW
BSFC,
kg/
hr
HC,
g/
hr
NOx,
g/
hr
CO,
kg/
hr
5785
4641
3485
2300
1725
778
453
225
76
5
262
159
97
54
17
3446
1519
2096
1374
384
7480
1842
351
84
34
115
105
100
66
9
weighted
result,
g/
kW­
hr
448
7.6
4.9
349
The
relatively
low
HC+
NOx
emissions
are
due
to
low
NOx
levels,
probably
due
to
a
rich
airfuel
ratio
and
cooled
charge
air.
HC
emissions
are
high,
but
within
the
upper
end
of
the
range
of
hydrocarbon
rates
for
typical
SD/
I
engines.
Rich
air­
fuel
ratios
likely
contribute
to
the
high
HC
and
CO
emissions.

This
engine
uses
electronically
controlled
fuel
injection
which
allows
the
engine
to
be
better
calibrated
for
more
precise
fuel
management.
However,
the
injectors
are
in
the
throttle
body
and
they
are
batch
fired,
four
at
a
time,
for
each
bank.
Therefore,
the
engine
still
must
be
calibrated
rich,
especially
at
idle,
to
avoid
lean
misfire
due
to
cylinder­
to­
cylinder
air­
fuel
ratio
variation.
Because
the
intake
air
is
supercharged,
cylinder­
to­
cylinder
variation
due
to
pressure
pulses
is
likely
minimal.
­
9­
Acknowledgments
We
would
like
to
thank
Jeff
Burrill
and
Tim
Cushing
at
Sterling
Performance
for
supplying
the
engines
and
the
testing
facility
as
well
as
for
their
time
and
expertise.
­
10­

Attachments
Attachment
1:
Detailed
Test
Data
for
Carbureted
Engine
Mode
Statistic
Engine
Speed
rpm
Power
hp
Fuel
Flow
lbs/
hr
Humidity
grains
/
lb.
dry
air
HC
ppm
NOx
corrected
ppm
CO
%
CO
2
%
Exhaust
A/
F
1
average
std.
dev.

COV
5802
3.9
0%
738
1.6
0%
362
2.1
1%
32.5
0.4
1%
7585
1101.1
15%
2982
703.8
24%
3.37
1.1
31%
12.78
1.5
12%
12.8
0.6
5%

2
average
std.
dev.

COV
4639
4.0
0%
424
1.0
0%
218
3.4
2%
32.7
0.7
2%
2476
36.1
1%
1181
17.6
1%
5.09
0.1
1%
12.13
0.0
0%
12.5
0.0
0%

3
average
std.
dev.

COV
3481
2.0
0%
205
1.3
1%
129
1.7
1%
33.8
0.4
1%
3058
30.0
1%
235
1.6
1%
9.28
0.1
1%
9.48
0.0
0%
11.2
0.0
0%

4
average
std.
dev.

COV
2321
4.2
0%
81
0.7
1%
65
2.0
3%
34.6
0.4
1%
6623
888.6
13%
115
1.3
1%
8.32
0.1
1%
9.92
0.1
1%
11.3
0.1
0%

5*
average
std.
dev.

COV
1345
9.9
1%
2
0.1
4%
19
1.3
7%
34.1
0.4
1%
11533
0.3
0%
269
1.2
0%
3.35
0.1
3%
8.45
0.1
2%
16.4
0.2
1%

*
Note:
The
HC
concentration
at
idle
appeared
to
be
above
the
highest
range
for
the
PEMS
unit.
­
11­

Attachment
2:
Detailed
Test
Data
for
Fuel­
Injected,
Supercharged
Engine
Mode
Statistic
Engine
Speed
rpm
Power
hp
Fuel
Flow
lbs/
hr
Humidity
grains
/
lb.
dry
air
HC
ppm
NOx
corrected
ppm
CO
%
CO
2
%
Exhaust
A/
F
1
average
std.
dev.

COV
5785
4.3
0%
1044
2.3
0%
577
7.2
1%
45.6
1.2
3%
1890
84.2
4%
1403
122.1
9%
3.56
0.3
8%
12.68
0.1
1%
13.3
0.2
1%

2
average
std.
dev.

COV
4641
1.8
0%
607
1.2
0%
351
2.1
1%
46.1
0.9
2%
1445
58.7
4%
605
28.2
5%
5.67
0.2
3%
11.65
0.1
1%
12.4
0.1
1%

3
average
std.
dev.

COV
3485
2.9
0%
301
0.8
0%
214
2.2
1%
47.2
1.0
2%
3472
368.0
11%
202
4.9
2%
9.51
0.4
5%
8.81
0.6
7%
11.4
0.1
1%

4
average
std.
dev.

COV
2300
5.1
0%
101
0.9
1%
118
1.3
1%
47.7
1.3
3%
4100
94.7
2%
87
0.8
1%
11.21
0.1
1%
6.88
0.0
0%
11.4
0.0
0%

5
average
std.
dev.

COV
1725
13.7
1%
7
0.2
3%
38
2.2
6%
47.8
1.0
2%
2753
165.6
6%
82
0.8
1%
3.58
0.2
5%
9.97
0.1
1%
15.6
0.1
1%
