Page
1
of
51
Test
Plan
to
Determine
the
PEMS
Measurement
Allowances
for
use
in
the
Manufacturer­
Run
Heavy­
Duty
Diesel
Engine
In­
Use
Testing
Program
Developed
by:

United
States
Environmental
Protection
Agency,
California
Air
Resources
Board,
Engine
Manufacturers
Association,
and
Southwest
Research
Institute
April
15,
2005
Page
2
of
51
Table
of
Contents
Executive
Summary
...................................................................................................................................
6
Introduction................................................................................................................................................
7
A­
1
Assessment
of
PEMS
measurement
errors,
under
standard
lab
conditions
(
PEMS
vs
LAB)..........
8
A­
1.1
Steady­
State
Testing.......................................................................................................................
8
A­
1.1.1
Goal:........................................................................................................................................
8
A­
1.1.2
Configuration:
.........................................................................................................................
8
A­
1.1.3
Guidelines:
..............................................................................................................................
8
A­
1.1.4
Anticipated
time
requirement
(
not
including
de­
bug
and
trouble­
shooting)
..............................
8
A­
1.1.4
Monte
Carlo
Modeling:
Emissions
Bias
and
Precision
Errors...................................................
9
A­
1.1.5
Monte
Carlo
Modeling:
Exhaust
flow­
rate
Bias
and
Precision
Errors.......................................
9
A­
1.2
Transient
Testing..........................................................................................................................
10
A­
1.2.1
Objective
...............................................................................................................................
10
A­
1.2.2
Background
...........................................................................................................................
10
A­
1.2.3
Methods
and
Materials...........................................................................................................
10
A­
1.2.4
Data
Analysis
........................................................................................................................
12
A­
2
PEMS
Exhaust
Flow
Meter
System
 
Sensor
Flow
Conditioning
Effects.......................................
14
A­
2.1
Purpose
........................................................................................................................................
14
A­
2.2
Basic
Approach
............................................................................................................................
14
A­
2.2.1
Pulsation
Effects
.............................................................................................................
14
A­
2.2.2
Non­
Uniform
Velocity
Effects
........................................................................................
14
A­
2.2.3
Tailpipe
Wind
Effects
.....................................................................................................
14
A­
2.3
Test
Configuration
Descriptions.............................................................................................
15
A­
2.3.1
Exhaust
Pulsation
Effects
................................................................................................
15
A­
2.3.2
Exhaust
Non­
Uniform
Velocity
Effects
...........................................................................
15
A­
2.3.3
Tailpipe
Wind
Effects
.....................................................................................................
15
A­
2.4
Engine
Test
Point
Description.................................................................................................
15
A­
2.5
Tailpipe
Wind
Effects
.............................................................................................................
16
A­
2.6
Data
Analyses
to
be
Performed
and
Reported..........................................................................
16
A­
3
Error
Assessment
of
ECM­
based
Torque
and
ECM­
based
BSFC
at
3rd
party
Engine
Lab.........
17
A­
3.1
Goal:
............................................................................................................................................
17
A­
3.2
Systems
and
Processes
to
be
used:
................................................................................................
17
A­
3.1
(
DOE
Testing):.............................................................................................................................
18
A­
3.2
(
Sensitivity
Testing):
....................................................................................................................
18
A­
3.3
(
Linearity
Error
Evaluation):
........................................................................................................
18
A­
3.4
Monte
Carlo
Model
......................................................................................................................
19
A­
3.4.1
Delta
Torque1:
Bias
and
variability
torque
errors
due
to
DOE
parameters..............................
19
A­
3.4.2
Delta
Torque2:
Torque
errors
due
to
changes
in
fuel
temperature...........................................
19
A­
3.4.3
Delta
Torque3:
Torque
errors
due
to
changes
in
oil
temperature.............................................
20
A­
3.4.4
Delta
Torque4:
Torque
errors
due
to
changes
in
coolant
temperature
.....................................
20
A­
3.4.5
Delta
Torque5:
Torque
errors
due
to
changes
in
humidity
......................................................
21
A­
3.4.6
Delta
Torque6:
Torque
errors
due
to
changes
in
fuel
properties..............................................
21
A­
3.4.7
Delta
Torque7:
Torque
errors
due
to
assumed
linearity
of
torque/
BSFC
map
.........................
22
Page
3
of
51
A­
3.4.8
Delta
Torque8:
Torque
errors
due
to
non­
deficiency
AECDs,
ECM
software
maps................
22
A­
3.4.9
Delta
Torque9:
Torque
errors
due
to
dynamic
measurement
(
instrument
response,
time
alignment)
..........................................................................................................................................
23
A­
3.4.10
Delta
Torque10:
Torque
errors
due
to
production
variability................................................
23
A­
3.4.11
Delta
Torque11:
Torque
errors
due
deterioration
factor........................................................
23
B­
1
Evaluation
of
the
effect
of
vehicle
shock,
vibration
and
orientation
on
the
performance
of
the
PEMS........................................................................................................................................................
25
B­
1.1
Objective
......................................................................................................................................
25
B­
1.2
Background
..................................................................................................................................
25
B­
1.3
Methods
and
Materials..................................................................................................................
25
B­
1.4
Quantifying
PEMS
performance
through
the
shock/
vibration
profiles...........................................
26
B­
2
Evaluation
of
ambient
temperature
effects
on
PEMS......................................................................
28
B­
2.1
Objective
......................................................................................................................................
28
B­
2.2
Background
..................................................................................................................................
28
B­
2.3
Methods
and
Materials..................................................................................................................
28
B­
2.4
Data
Analysis
...............................................................................................................................
30
B­
3
Evaluation
of
ambient
pressure
effects
on
PEMS............................................................................
31
B­
3.1
Objective
......................................................................................................................................
31
B­
3.2
Background
..................................................................................................................................
31
B­
3.3
Methods
and
Materials..................................................................................................................
31
B­
3.4
Data
Analysis
...............................................................................................................................
33
B­
4
Evaluation
of
the
effect
of
Electromagnetic
Interference
and
Radio
Frequency
Interference
on
the
performance
of
the
PEMS
.......................................................................................................................
34
B­
4.1
Objective
......................................................................................................................................
34
B­
4.2
Background
..................................................................................................................................
34
B­
4.3
Methods
and
Materials..................................................................................................................
35
B­
4.3.1
Radiated
Immunity:
...............................................................................................................
35
B­
4.4
Quantifying
PEMS
performance
through
the
EMI/
RFI
evaluation:
...............................................
36
B­
5
Evaluation
of
ambient
hydrocarbons
effects
on
PEMS
...................................................................
38
B­
5.1
Objective
......................................................................................................................................
38
B­
5.2
Background
..................................................................................................................................
38
B­
5.3
Methods
and
Materials..................................................................................................................
38
B­
5.4
Data
Analysis
...............................................................................................................................
42
B­
6
Evaluation
of
baseline
repeatability
and
bias
of
PEMS...................................................................
43
B­
6.1
Objective
......................................................................................................................................
43
B­
6.2
Background
..................................................................................................................................
43
B­
6.3
Methods
and
Materials..................................................................................................................
43
B­
6.4
Data
Analysis
...............................................................................................................................
44
C­
1
Error
Assessment
of
ECM­
based
Torque
and
ECM­
based
BSFC,
at
Engine
Manufacturers
individual
laboratories
.............................................................................................................................
45
Page
4
of
51
C­
1.1
Goal..............................................................................................................................................
45
C­
1.2
Systems
and
Processes
to
be
used:
................................................................................................
45
C­
1.3
Data
Analysis
...............................................................................................................................
45
C­
1.4
Deadline
.......................................................................................................................................
45
D­
1
Monte
Carlo
Error
Model
and
Measurement
Allowance................................................................
46
D­
1.1
Objective......................................................................................................................................
46
D­
1.2
Overview......................................................................................................................................
46
D­
1.3
Methods
and
Materials
.................................................................................................................
47
D­
1.4
Data
Analysis
...............................................................................................................................
47
E­
1
Validation
of
the
Monte
Carlo
Model...............................................................................................
48
E­
1.1
Objective
......................................................................................................................................
48
E­
1.2
Background
..................................................................................................................................
48
E­
1.3
Methods
and
Materials..................................................................................................................
48
E­
1.4
Data
Analysis................................................................................................................................
49
Page
5
of
51
Table
of
Figures
and
Tables
Table
1:
30
different
32­
s
NTE
events
.......................................................................................................
11
Table
2:
Inter­
NTE
event
Intervals1
...........................................................................................................
12
Table
3:
Gas
cylinder
contents...................................................................................................................
26
Table
4:
Time
and
temperature
sequence
for
8­
hour
evaluation
of
ambient
temperature
effects
.................
28
Figure
1:
Temperature
Time
Profile
for
PEMS
Testing.
............................................................................
29
Table
5:
Gas
cylinder
contents...................................................................................................................
29
Table
6:
Time
and
pressure
sequence
for
8­
hour
evaluation
of
ambient
pressure
effects
............................
31
Figure
2:
Pressure­
Time
Profile
for
PEMS
Testing....................................................................................
32
Table
7:
Gas
cylinder
contents...................................................................................................................
32
Table
8:
Gas
cylinder
contents...................................................................................................................
36
Table
9:
Hydrocarbons
sequence
for
evaluation
of
ambient
hydrocarbons
effects
......................................
39
Table
10:
Gas
cylinder
contents.................................................................................................................
41
Table
11:
Gas
cylinder
contents.................................................................................................................
43
Table
12:
Monte
Carlo
Model
Validation
..................................................................................................
49
Figure
3:
Model
Validation
NTE
Events
...................................................................................................
50
Figure
4:
PEMS
vs
Ce­
Cert
Brake­
Specific
Data.......................................................................................
51
Page
6
of
51
Executive
Summary
Page
7
of
51
Introduction
Page
8
of
51
A­
1
Assessment
of
PEMS
measurement
errors,
under
standard
lab
conditions
(
PEMS
vs
LAB)

A­
1.1
Steady­
State
Testing
A­
1.1.1
Goal:

a)
Obtain
data
for
torque
mapping
and
test
matrix
selection
(
40
data
points)
b)
Quantify
bias
and
precision
errors
of
gaseous
emissions
concentration
measurements
(
ppm
raw
and
fuel­
specific
dilute),
under
standard
laboratory
conditions.
c)
Quantify
bias
and
precision
errors
in
exhaust
flow
measurement
measurements
using
PEMS'
portable
flow­
meters,
under
standard
laboratory
conditions.
d)
Quantify
bias
and
precision
errors
in
gaseous
emissions
flow
rate
measurements
(
e.
g.
g/
hr
NOx).

A­
1.1.2
Configuration:

During
this
task,
the
following
systems
will
be
evaluated:
a)
Three
(
3)
heavy
duty
diesel
engines
(
1
HHDE,
1
MHDE,
1
LHDE)
b)
Six
(
6)
PEMS
analyzers
(
3
Sensors
Semtech­
D
models,
and
3
Horiba
OBS­
2
models)
c)
Six
(
6)
PEMS
exhaust
flow­
meters
(
3
from
Sensors,
and
3
from
Horiba)
d)
One
third­
party
engine
laboratory
will
be
selected
and
used
for
all
testing
in
this
task.

A­
1.1.3
Guidelines:

1.
Measure
raw
as
well
as
CVS­
dilute
emissions
2.
Measure
engine
inlet
airflow
through
use
of
LFE
or
equivalent
3.
Measure
fuel
consumption
and
torque
4.
No
need
to
measure
PM?
(
time
and
$).
Just
measure
AVL
smoke?
5.
Steady­
state
testing
6.
Stabilization
time
=
90
seconds.
Data
acquisition
=
30
seconds,
after
stabilization.
Dwell
time
between
points
=
30
seconds
(
total
time
per
point
=
150
sec.
=
2.5
min)
7.
Measure
40
points
spanning
the
NTE
zone
(
no
repeats.
This
is
for
mapping
only)
8.
Down
select
10
operating
conditions
from
40
points,
to
be
used
during
repeat
testing.
9.
Repeat
tests:
20
times
per
point
10.
Repeat
testing
will
change
the
order
of
testing
from
run
to
run
(
first
data
pt
in
set
one
will
not
be
first
during
second
test,
etc.)
11.
Selection
of
10
data
points
should
appropriately
span
range
of
expected
emissions
concentrations
and
exhaust
flowrate
(
obtained
during
40
point
test
set)
12.
Each
test
will
use
one
PEMS
manufacturer
at
a
time,
to
measure
emissions
concentration
and
exhaust
flowrate
(
unless
otherwise
agreed
to
by
PEMS
manufacturer)
13.
Expected
test
duration:
10
points
x
2.5
minutes
x
20
repeats
=
500
minutes
(
8.3
hrs)
14.
Each
engine
will
have
to
be
run
on
6
different
days
to
obtain
data
from
all
6
PEMS
A­
1.1.4
Anticipated
time
requirement
(
not
including
de­
bug
and
trouble­
shooting)

e)
Mapping:
3
days
total
(
1
day/
engine)
f)
Bias
and
Precision
assessment:
18
days
(
6
days/
engine)
Page
9
of
51
A­
1.1.4
Monte
Carlo
Modeling:
Emissions
Bias
and
Precision
Errors
a)
 (
MEAS)
will
be
the
"
delta"
to
be
used
in
the
Monte
Carlo
model
to
assess
the
variability
in
emissions
concentration
due
to
steady
state
bias
and
precision
errors
b)
Data
will
first
be
plotted
as
follows:
i.
x­
axis:
LAB
data
(
ppm
for
raw,
or
fuel­
specific
%
for
dilute
measurements)
ii.
y­
axis:
 (
MEAS)
=
PEMS
data
minus
LAB
data
(
ppm
or
fuel­
specific
%)
c)
 (
MEAS)
will
then
be
plotted
in
a
3­
D
chart
as
the
z­
axis
against:
i.
x­
axis:
LAB
data
(
ppm
or
raw)
ii.
variability
index
"
ic"
(
range
 
1
to
+
1
)
d)
The
variability
index
"
ic"
will
in
essence
be
the
dice
to
roll
in
the
Monte
Carlo
simulation
e)
Probability
distribution
for
"
ic"
will
likely
be
forced
to
be
Gaussian
f)
The
variability
index
"
ic"
will
change
during
each
step
of
the
integration
scheme
in
the
Monte
Carlo
simulation
(
i.
e.,
not
constant
throughout
the
NTE
event).

A­
1.1.5
Monte
Carlo
Modeling:
Exhaust
flow­
rate
Bias
and
Precision
Errors
g)
 (
MEAS)
will
be
the
"
delta"
to
be
used
in
the
Monte
Carlo
model
to
assess
the
variability
in
emissions
concentration
due
to
steady
state
bias
and
precision
errors
h)
Data
will
first
be
plotted
as
follows:
i.
x­
axis:
LAB
data
(
kg/
hr)
ii.
y­
axis:
 (
MEAS)
=
PEMS
exhaust
flow
minus
LAB
exhaust
flow
(
kg/
hr)
i)
 (
MEAS)
will
then
be
plotted
in
a
3­
D
chart
as
the
z­
axis
against:
i.
x­
axis:
LAB
data
(
kg/
hr)
ii.
variability
index
"
ic"
(
range
 
1
to
+
1
)
j)
The
variability
index
"
ic"
will
in
essence
be
the
dice
to
roll
in
the
Monte
Carlo
simulation
k)
Probability
distribution
for
"
ic"
will
likely
be
forced
to
be
Gaussian
l)
The
variability
index
"
ic"
will
change
during
each
step
of
the
integration
scheme
in
the
Monte
Carlo
Note
1:
engines
used
will
have
to
be
compliant
with
EPA's
current
standard
emissions
regulations.
EPA
has
requested
that
diesel
particulate
filters
(
DPF)
are
fitted
to
these
engines.
Note
2:
more
discussion
is
needed
to
assess
long
term
variability
Page
10
of
51
A­
1.2
Transient
Testing
A­
1.2.1
Objective
Evaluate
the
repeatability
of
measuring
30
different
32­
second
NTE
events
in
different
orders
over
a
20­
minute
cycle.

A­
1.2.2
Background
PEMS
are
expected
to
operate
in
a
repeatable
manner
over
NTE
events
as
short
as
30
seconds.
It
is
hypothesized
that
two
sources
of
PEMS
repeatability
error
may
a
PEMS'
dynamic
response
to
rapidly
changing
signals
and
its
susceptibility
to
"
history"
effects.
Dynamic
response
error
includes
error
due
to
measurement
signal
time
alignment,
and
the
dissimilarity
of
the
dynamic
response
and
aliasing
of
signals
that
are
combined
on
a
second­
by­
second
basis;
including
those
signals
used
to
determine
entry
into
and
exit
from
the
NTE
zone.
History
effects
include
the
effects
of
previously
measured
quantities
on
currently
measured
quantities.
For
example,
this
may
be
caused
by
ineffective
sample
exchange
in
the
gaseous
emissions
sampling
volumes,
or
it
may
be
caused
by
one
or
more
sensors'
characteristic
rise
time
and
fall
time.
To
account
for
any
dynamic
response
repeatability
error
the
increase
in
repeatability
error
incremental
to
steady­
state
emissions
measurement
repeatability
will
be
incorporated
into
the
overall
error
model.

A­
1.2.3
Methods
and
Materials
For
this
experiment
an
engine
dynamometer
emissions
test
cell
is
needed.
The
test
cell
area
must
be
able
to
accommodate
at
least
six
PEMS,
their
power
supplies,
the
PEMS
flow
meters,
cables
and
lines.

This
experiment
will
challenge
PEMS
to
30
different
32­
second
NTE
events
over
a
20
minute
test
cycle.
Selection
of
the
short
NTE
cycles
(
all
32­
seconds)
maximizes
the
sensitivity
of
this
test
to
effects
of
dynamic
response.
Thirty­
two
seconds
was
chosen
as
the
minimum
instead
of
thirty
seconds
(
which
is
the
shortest
NTE
event
time).
This
was
done
to
ensure
that
1
Hz
ECM
updating
of
torque
and
speed
values
would
be
unlikely
to
interfere
with
capturing
NTE
events.
For
each
repeat
of
the
test
cycle,
the
order
of
the
30
different
NTE
events
will
be
randomly
rearranged.
In
addition
the
29
different
intervals
separating
each
NTE
event
from
the
next
will
have
a
range
of
durations
and
these
that
will
be
randomly
arranged
in
the
test
cycle
as
well.
Random
rearrangement
of
the
NTE
events
and
inter­
NTE
events
will
maximize
the
sensitivity
of
this
test
to
dynamic
response
and
history
effects.
Page
11
of
51
Table
1:
30
different
32­
s
NTE
events
NTE
Event
1Speed
%
Range
2Torque
%
Range
Description
NTE1
17%
332%
Steady
speed
and
torque;
lower
left
of
NTE
NTE2
(
17%+
MTS)/
2
332%
Steady
speed
and
torque;
lower
center
of
NTE
NTE3
Governor
line
3
32%
Steady
speed
and
torque;
lower
right
of
NTE
NTE4
17%
(
NTE1+
NTE7)/
2
Steady
speed
and
torque;
middle
left
of
NTE
NTE5
(
17%+
MTS)/
2
(
NTE2+
NTE8)/
2
Steady
speed
and
torque;
middle
center
of
NTE
NTE6
Governor
line
(
NTE3+
NTE9)/
2
Steady
speed
and
torque;
middle
right
of
NTE
NTE7
17%
100%
Steady
speed
and
torque;
upper
left
of
NTE
NTE8
(
17%+
MTS)/
2
100%
Steady
speed
and
torque;
upper
center
of
NTE
NTE9
MTS
100%
Steady
speed
and
torque;
upper
right
of
NTE
NTE10
Lower
third
332%
­
100%
Highly
transient
torque;
moderate
transient
speed
NTE11
Upper
third
332%
­
100%
Highly
transient
torque;
moderate
transient
speed
NTE12
Middle
third
332%
­
100%
Highly
transient
torque;
moderate
transient
speed
NTE13
17%
­
governed
Lower
third
Highly
transient
speed;
moderate
transient
torque
NTE14
17%
­
governed
Upper
third
Highly
transient
speed;
moderate
transient
torque
NTE15
17%
­
governed
Middle
third
Highly
transient
speed;
moderate
transient
torque
NTE16
Lower
right
diagonal
Transient;
speed
increases
as
torque
increases
NTE17
Upper
left
diagonal
Transient;
speed
increases
as
torque
increases
NTE18
Full
diagonal;
lower
left
to
upper
right
Transient;
speed
increases
as
torque
increases
NTE19
Lower
left
diagonal
Transient;
speed
decreases
as
torque
increases
NTE20
Upper
right
diagonal
Transient;
speed
decreases
as
torque
increases
NTE21
Full
diagonal;
lower
right
to
upper
left
Transient;
speed
decreases
as
torque
increases
NTE22
Cruise;
~
60
mph
Sample
from
HDDE
NTE23
Cruise;
~
50
mph
Sample
from
HDDE
NTE24
Cruise;
~
75
mph
Sample
from
HDDE
NTE25
Small
bulldozer
Sample
from
NRDE
NTE26
Large
bulldozer
Sample
from
NRDE
NTE27
Second
of
three
NTE
events
in
FTP
Seconds
used
from
FTP:
714­
725,
729­
743,
751­
755
NTE28
Third
of
three
NTE
events
in
FTP
Seconds
used
from
FTP:
790­
797,
799­
802,
805­
810,
813­
819,
821­
826,
831
NTE29
First
of
two
NTE
events
in
NRTC
Seconds
used
from
NRTC:
423­
430,
444,
448­
450,
462­
481,
increased
464
speed
from
40%
to
42%
NTE30
First
of
two
NTE
events
in
NRTC
Seconds
used
from
NRTC:
627­
629,
657­
664,
685­
696,
714­
722
1
Speed
(
rpm)
=
Curb
Idle
+
(
Speed
%
*
(
MTS
­
Curb
Idle)
2
Torque
(
lbf­
ft)
=
Torque
%
*
Maximum
Torque
At
Speed
(
i.
e.
lug
curve
torque
at
speed)
3
Torque
(
lbf­
ft)
=
Maximum
of
(
32
%
*
peak
torque)
and
the
torque
at
speed
that
produces
(
32
%
*
peak
power)
Page
12
of
51
Table
2:
Inter­
NTE
event
Intervals1
INT
Event
Duration
(
s)
Frequency
Description
INT1
10
1
Initiation
of
cycle;
INT1
is
always
first
INT2­
6
2
5
Shortest
and
most
frequent
inter­
NTE
events
INT7­
10
3
4
Short
and
frequent
inter­
NTE
events
INT11­
14
4
4
Short
and
frequent
inter­
NTE
events
INT15­
18
5
4
Short
and
frequent
inter­
NTE
events
INT19­
21
6
3
Short
and
frequent
inter­
NTE
events
INT22
7
1
Medium
inter­
NTE
event
INT23
8
1
Medium
inter­
NTE
event
INT24
9
1
Medium
inter­
NTE
event
INT25
11
1
Medium
inter­
NTE
event
INT26
13
1
Long
inter­
NTE
event
INT27
17
1
Long
inter­
NTE
event
INT28
22
1
Long
inter­
NTE
event
INT29
27
1
Long
inter­
NTE
event
INT30
35
1
Longest
inter­
NTE
event
INT31
5
1
Termination
of
cycle;
INT31*
is
always
last
1Interval
speeds
and
torques
are
not
identical,
but
they
are
clustered
around
the
operating
point
at
which
15%
of
peak
power
and
15%
of
peak
torque
are
output.

Prior
to
executing
the
first
cycle,
each
PEMS
will
be
setup,
and
the
engine
will
be
stabilized
at
the
first
inter­
NTE
event
operating
point
described
in
Table
?.
The
PEMS
setup
will
be
conducted
according
to
PEMS
manufacturer
instructions,
including
any
warm­
up
time,
zero­
spans
of
the
analyzers
and
the
setup
of
all
accessories
including
flow
meters,
ECM
interpreters,
etc.
Then,
when
the
test
cycle
starts,
the
PEMS'
will
be
switched
to
sample
emissions
from
the
engine.
When
the
text
cycle
ends,
the
PEMS'
will
be
switched
back
to
ambient
sampling,
and
all
post­
test
PEMS
validations
will
be
completed
according
to
PEMS
manufacturer
instructions.

This
test
cycle
will
be
repeated
20
times
for
a
total
of
600
measured
NTE
events.
The
test
cycle
will
be
repeated
4
to
5
times
per
day
and
over
4
to
5
different
days.

A­
1.2.4
Data
Analysis
NTE
events
invalidated
by
any
criteria
in
40
CFR
Part
1065
will
be
discarded
from
further
analysis.
For
each
NTE
event,
brake­
specific
means
for
NOx,
NMHC,
and
CO
will
be
calculated
using
all
of
the
valid
repeats
of
that
event
using
the
three
different
work
calculation
methods.
The
three
methods
are
given
below,
using
NOx
as
an
example:
Page
13
of
51
2
2
2
1
1
1
1
1
1
(
g/
hp­
hr)

(
g/
hp­
hr)

1
(
1
)

(
g/
hp­
hr)
x
x
x
N
NO
i
i
i
NO
N
ni
i
i
N
NO
i
i
i
NO
Cproddryi
i
N
C
H2Oi
i
fuel
fueli
N
NO
fuel
fueli
H2Oi
i
i
C
Cproddryi
NO
N
ni
i
i
M
xn
t
e
f
T
t
M
xn
t
e
x
n
M
x
t
w
e
M
w
m
x
x
t
M
x
e
f
T
=

=

=

=

=

=
D
=
D
D
=

+
D
+
D
=
å
å
å
å
å
&
g
g
g
g
g
%&
g
g
g
%&
g
g
g
&
g
g
g
g
g
g
1
record
t
t
f
D
D
=
å
g
Then
for
each
event
and
calculation
method,
differences
between
the
individual
values
and
their
respective
means
will
be
calculated.
Next
decrease
the
magnitude
of
each
difference
by
the
corresponding
steady­
state
repeatability
difference,
which
was
determined
in
the
steady­
state
repeatability
and
bias
evaluation.
It
is
possible
that
this
adjustment
will
decrease
some
difference
magnitudes
to
zero.
The
end
result
will
be
distributions
of
differences
(
including
any
zero
magnitude
differences)
for
each
NTE
event
and
calculation
method.
In
the
error
model,
for
each
modeled
brake­
specific
NTE
result
(
and
for
each
calculation
method),
the
respective
distribution
of
differences
with
the
mean
value
closest
to
the
model's
brake­
specific
value
will
be
randomly
sampled,
and
the
sampled
value
will
be
added
(
or
subtracted,
depending
upon
the
sign
of
the
sampled
value)
to
the
model's
brake­
specific
NTE
result.
This
incorporates
into
the
model
the
effect
of
dynamic
response
repeatability.

Note
that
a
new
sampled
difference
should
be
chosen
once
per
NTE
event
because
it
was
determined
once
per
NTE
event,
not
once
per
second.

Note
too
that
in
order
to
best
incorporate
the
data
from
this
experiment
into
the
model,
the
model
must
have
nominal
brake­
specific
values
spread
throughout
the
range
of
the
mean
NTE
values
generated
in
this
experiment.
This
will
ensure
that
the
distributions
from
many
of
the
NTE
events
measured
in
this
experiment
will
be
incorporated
into
the
model.
Page
14
of
51
A­
2
PEMS
Exhaust
Flow
Meter
System
 
Sensor
Flow
Conditioning
Effects.

A­
2.1
Purpose
The
purpose
of
this
test
program
is
to
extend
the
evaluation
of
the
PEMS
exhaust
flow
metering
system
established
in
Section
A.
1
by
evaluating
the
sensitivity
of
the
exhaust
flow
sensor
to
exhaust
installation
related
factors
such
as
pulsating
exhaust
flows,
exhaust
piping
configuration
influences
on
the
sensor,
and
possible
wind
effects
on
the
exhaust
tailpipes.
Section
A.
1
focused
on
the
accuracy
and
linearity
characteristics
of
the
flow
sensors
and
provides
the
baseline
data
for
quantifying
the
flow
metering
effects
identified
during
this
testing.

A­
2.2
Basic
Approach
The
basic
approach
of
this
test
program
is
to
expose
the
PEMS
flow
sensors
to
exhaust
flow
conditions
that
could
adversely
affect
the
velocity
profiles
surrounding
the
flow
sensors
in
terms
of
exhaust
pulsations,
nonuniform
velocity
profiles,
and
wind
velocity
effects
on
the
tailpipe
exit.

A­
2.2.1
Pulsation
Effects
The
PEMS
exhaust
flow
sensors
are
to
be
exposed
to
engine
exhaust
pulsations
by
repeating
the
baseline
tests
conducted
in
Section
A.
1,
with
the
exception
that
the
aftertreatment
device
is
removed
from
the
exhaust
system
and
replaced
with
an
unobstructed
full
flow
exhaust
pipe.
Without
the
pulsation
dampening
effect
of
the
aftertreatment
device,
the
flow
sensors
will
be
exposed
to
higher
pulsations
than
found
in
typical
exhaust
systems.

The
accuracy
and
linearity
test
data
from
Section
A.
1
provides
the
baseline
information
in
which
to
compare
the
flow
sensor
responses
with
and
without
the
pulsation
attenuating
effects
of
the
exhaust
aftertreatment
device.
Flow
deviations
from
the
Section
A.
1
baseline
data
are
an
indication
of
flow
pulsations
affecting
the
PEMS
reported
flow
rate.

A­
2.2.2
Non­
Uniform
Velocity
Effects
For
this
test
phase
the
effect
of
piping
elbows
that
may
be
used
to
connect
the
PEMS
exhaust
flow
sensor
to
a
vehicle
exhaust
system
are
to
be
evaluated.
This
is
to
be
done
by
installing
two
90
º
elbows
in
series,
in
non­
parallel
planes,
immediately
upstream
of
the
exhaust
flow
metering
device
supplied
by
the
PEMS
manufacturer.
The
intent
of
the
two
exhaust
elbows
in
non­
parallel
planes
is
to
induce
swirl
into
the
exhaust
flow
which
produces
non­
uniform
flow
velocity
profiles
across
the
exhaust
pipe
prior
to
the
PEMS
flow
metering
system.

The
baseline
flow
evaluation
data
of
Section
A.
1
provides
the
baseline
flow
data
in
which
to
compare
the
PEMS
exhaust
flow
meter
response
with
the
additional
piping
elbows
installed.

A­
2.2.3
Tailpipe
Wind
Effects
When
installed
on
a
test
vehicle,
the
exhaust
outlet
of
the
PEMS
(
the
tailpipe)
will
be
exposed
to
air
currents
that
could
adversely
affect
the
metering
characteristics
of
the
exhaust
flow
sensor.
In
as
much
as
these
effects
will
likely
be
somewhat
proportional
to
vehicle
speed,
it
would
be
very
difficult
to
quantify
these
Page
15
of
51
effects
in
a
meaningful
manner
as
air
current
velocities
and
directions
around
a
vehicle
cab
could
be
complex.

However,
it
is
felt
that
some
effort
is
justified
to
expose
the
exhaust
tailpipe
to
high
velocity
air
movement
from
various
directions
and
angles
and
noting
any
deviation
from
baseline
flow
measurements.
If
flow
deviations
are
noted
during
this
test
phase,
this
topic
may
warrant
additional
exploration
and/
or
quantification
of
the
air
velocities
utilized
during
the
testing.

A­
2.3
Test
Configuration
Descriptions
Evaluation
of
the
PEMS
exhaust
flow
sensors
for
the
section
(
Section
A.
2)
must
be
performed
after
the
Section
A.
1
testing
program,
as
the
accuracy
and
linearity
data
from
the
Section
A.
1
program
provides
the
baseline
data
for
comparison
to
the
data
generated
for
this
section.

A­
2.3.1
Exhaust
Pulsation
Effects
The
baseline
data
for
comparing
the
exhaust
pulsation
effects
on
the
flow
sensors
will
be
the
Section
A.
1
data
with
the
exhaust
aftertreatment
device
installed.
As
such,
the
effect
of
exhaust
pulsations
test
is
to
be
conducted
with
the
same
exhaust
configuration,
with
the
exception
that
the
aftertreatment
device
is
to
be
removed
and
replaced
with
a
section
of
unobstructed
exhaust
pipe
of
the
same
diameter
as
pipes
entering
and
exiting
the
aftertreatment
device.

With
the
open
exhaust
pipe
installed,
the
engine
is
to
be
run
in
such
a
manner
that
the
same
exhaust
flow
rates
are
achieved,
as
determined
by
the
Laboratory
system,
as
used
in
Section
A.
1
to
quantify
the
accuracy
and
linearity
characteristics
of
the
flow
sensor.
Any
observed
change
in
the
Laboratory­
to­
PEMS
exhaust
flow
rate
relationship
from
the
Section
A.
1
data
should
be
attributed
to
flow
sensor
pulsation
sensitivity.

A­
2.3.2
Exhaust
Non­
Uniform
Velocity
Effects
As
with
exhaust
pulsations
effects
test
(
section
A.
2.3.1,
above),
the
same
test
data
from
Section
A.
1
will
be
used
as
the
baseline
data
for
comparison
to
the
data
generated
for
this
test.
In
order
to
keep
the
pulsation
effects
test
data
separate
from
the
non­
uniform
velocity
effects
test
(
this
test),
the
same
exhaust
system
and
aftertreatment
device
used
in
Section
A.
1
is
to
be
used
for
this
testing
phase.

The
two
90
º
pipe
elbows
are
to
be
installed
at
the
end
of
the
vehicle
exhaust
system,
immediately
upstream
of
the
PEMS
exhaust
flow
metering
apparatus.
The
two
elbows
are
to
be
connected
together
in
a
manner
such
that
their
axial
planes
are
90
º
from
each
other.
The
intent
of
this
set­
up
is
to
induce
swirl
into
the
exhaust
velocity
profile
at
a
point
just
prior
to
the
flow
entering
the
PEMS
flow
metering
system.

A­
2.3.3
Tailpipe
Wind
Effects
For
this
test
phase,
the
exhaust
system
set­
up
used
in
Section
A.
1
that
includes
the
exhaust
aftertreatment
device,
is
to
be
used
for
this
phase
of
testing.
As
detailed
in
section
A.
2.5,
high
velocity
air
is
to
be
blown
across
the
tailpipe
exit
in
an
attempt
to
see
if
the
indicated
exhaust
flow
rate
shows
any
significant
changes.
The
intent
of
this
test
phase
is
not
to
quantify
wind
effects,
but
only
to
determine
if
the
probability
exists
that
the
tailpipe
may
be
sensitive
to
wind
effects.

A­
2.4
Engine
Test
Point
Description
Page
16
of
51
The
engine
operating
conditions
and
engine
speed
and
loads
used
to
evaluate
the
exhaust
pulsation
effects
(
section
A.
2.3.1)
and
the
non­
uniform
exhaust
velocity
effects
(
section
A.
2.3.2)
are
to
follow
the
same
NTE
test
points
as
utilized
in
Section
A.
1
for
determining
the
accuracy
and
linearity
of
the
PEMS
flow
sensors.

As
these
effects
need
to
be
quantified
for
inclusion
in
the
PEMS
margin
analysis,
the
number
of
test
repeats
should
mirror
the
number
of
repeats
utilized
in
Section
A.
1.
However,
if
initial
test
data
indicates
these
effects
to
be
non­
existent,
then
a
smaller
data
set
can
be
considered.

A­
2.5
Tailpipe
Wind
Effects
The
intent
of
this
test
phase
is
to
assess
if
the
PEMS
exhaust
flow
metering
system
shows
any
sensitivity
to
high
air
velocities
in
the
area
of
the
exhaust
tailpipe
exit.
It
is
possible
that
the
exhaust
exit
may
be
the
exit
of
the
PEMS
exhaust
flow
metering
apparatus.
Regardless
of
the
type
of
exhaust
exit,
it
is
likely
that
during
actual
in­
use
tests
the
exhaust
exit
will
be
exposed
to
high
air
velocities
that
may
approach
the
exhaust
exit
from
relatively
unpredictable
directions,
and
that
flow
disruptions
at
the
PEMS
tailpipe
exit
may
adversely
affect
the
flow
metering
of
the
sensor.
It
should
be
noted
that
this
test
plan
in
not
designed
to
quantify
any
velocity
effects,
but
to
assess
if
a
more
sophisticated
approach
is
justified.
The
initial
assumption
is
that
these
effects
are
negligible.

The
approach
is
to
utilize
an
exhaust
test
set­
up
in
which
the
exhaust
can
freely
exit
the
PEMS
flow
metering
device
and
form
a
typical
exhaust
plume.
With
the
engine
operating
at
the
exhaust
flow
rates
established
in
Section
A.
1
(
accuracy
and
linearity
baseline
data)
direct
high
velocity
air
at
the
tailpipe
exit
from
a
plane
perpendicular
to
the
axial
exit
of
the
exhaust.
It
is
suggested
that
the
high
velocity
air
could
be
generated
via
a
common
leaf
blower
type
apparatus.
In
addition,
some
test­
points
are
to
be
made
directing
the
high
velocity
air
at
the
tailpipe
exit
from
other
than
perpendicular
to
the
exhaust
flow.

Should
the
PEMS
exhaust
flow
sensor
show
sensitivity
to
the
high
velocity
air
currents
by
indicating
a
flow
rate
change
compared
to
the
Section
A.
1
baseline
data,
or
additional
signal
noise
becomes
apparent
in
the
PEMS
indicated
exhaust
flow
rate,
this
information
is
to
be
noted.
In
such
a
case,
discussions
must
take
place
to
determine
if
a
more
quantitative
type
test
plan
is
warranted.

A­
2.6
Data
Analyses
to
be
Performed
and
Reported
a.
Comparison
tables/
charts/
comments
showing
the
influence
of
exhaust
pulsations
on
PEMS
flow
metering
accuracy.
The
data
comparisons
are
to
be
made
between
the
PEMS
indicated
flow
rate
and
the
Laboratory
indicated
flow
rate.

b.
Comparison
tables/
charts/
comments
describing
the
PEMS
exhaust
system
mounting
system
and
its
sensitivity
to
measurement
error
due
to
non­
uniform
flow
velocities
in
the
exhaust
pipe.

c.
Statements
concerning
observations
on
the
effect
of
high
velocity
air
currents
directed
at
the
exhaust
tailpipe.
Page
17
of
51
A­
3
Error
Assessment
of
ECM­
based
Torque
and
ECM­
based
BSFC
at
3rd
party
Engine
Lab
A­
3.1
Goal:
The
objective
of
this
task
is
to
quantify,
at
a
third
party
engine
lab,
errors
associated
with
"
estimating"
engine
torque
and
BSFC
through
ECM­
based
parameters
(
speed,
fuel
commanded).
Additional
errors
will
be
evaluated
at
engine
manufacturer
labs
(
involving
testing
that
would
be
considered
confidential
for
each
engine
manufacturer),
and
will
be
covered
under
a
different
task.

Note
1:
The
accuracy
of
the
torque/
bsfc
maps
is
only
relevant
to
in­
use
emissions
testing
when
under
the
NTE
zone,
and
in
operating
conditions
not
declared
as
deficiencies.
Note
2:
It
is
not
the
intent
of
this
task
to
"
minimize"
bsfc/
torque
mapping
errors
by
developing
more
sophisticated
mapping
techniques
(
that
would
impose
new
demands
in
normal
engine
development
processes,
for
every
engine
rating).
Note
3:
EPA
has
requested
to
note
that
if
data
shows
that
excessive
excursions
in
estimated
BSFC/
Torque
from
nominal
are
recorded
in
isolated
cases
(
when
operating
under
non­
deficiency
declared
NTE
zones
(
AECDs))
they
may
request
engine
manufacturers
not
to
include
those
conditions
as
part
of
the
measurement
allowance,
but
rather
explain
and
if
necessary
correct
these
"
isolated"
instances.

A­
3.2
Systems
and
Processes
to
be
used:
During
this
task,
two
different
sets
of
parameters
will
be
used
to
evaluate
their
impact
on
bsfc/
torque
map
errors.
The
first
one
will
use
parameters
that
are
likely
to
interact
with
each
other,
and
thus
a
Design
of
Experiments
(
DOE)
approach
will
be
used.
The
second
subtask
will
use
a
simpler
testing
approach,
and
will
use
parameters
that
are
not
likely
to
have
strong
interactions.
The
parameters
to
be
used
for
the
investigation
of
their
corresponding
effect
on
torque/
bsfc
map
accuracy
are
listed
below:

Subtask
A­
3.1:
will
evaluate
the
effect
of
the
following
parameters
on
bsfc/
torque
map
errors:
(
1)
intake
air
restriction;
(
2)
exhaust
gas
restriction;
(
3)
barometric
pressure
(
altitude);
(
4)
IMT
(
controlled
through
aftercooler
water
inlet
temperature/
flow;
only
for
operating
conditions
not
declared
as
deficiencies)
Subtask
A­
3.2:
will
evaluate
the
effect
of
the
following
parameters
on
bsfc/
torque
map
errors:
(
1)
oil
viscosity
(
weight);
(
2)
fuel
temperature
(??
Is
there
much
variation
in
this??);
(
3)
oil
temperature
(
can
investigate
either
through
engine
warm­
up
(
recording
data
from
when
the
engine's
operating
conditions
meet
criteria
for
a
valid
NTE
up
until
engine
is
fully
warm
OR
run
the
oil
from
the
sump
to
a
cooler
to
control
temperatures
(
4)
engine
coolant
temperature;
(
5)
intake
air
humidity;
(
6)
fuel
properties
(
cetane
number,
viscosity,
API
density;
possibly
investigate
with
a
D1
&
D2
fuel
or
with
2­
3
D2
fuels
with
different
properties)

More
discussion
is
still
needed.
For
time
being
group
agreed
to
investigate
the
temperature
effects
(
1­
4)
under
engine
warm
up,
recoding
data
from
when
the
engine's
operating
conditions
meet
criteria
for
a
valid
NTE
up
until
engine
is
fully
warm).
The
Cat
engine
(
no
EGR)
will
be
run
in
a
cold
cell
at
0
deg.
C
ambient
until
coolant
temp.
is
212F
+.
EGR
engines
will
start
test
at
ambient
temp.
Issues
(
5)
and
(
6)
will
be
investigated
separately
through
a
sensitivity
study.

Subtask
A­
3.3:
will
evaluate
the
effect
interpolating
the
torque/
BSFC
maps
off
of
the
down
sampled
20
SS
points
that
are
used
to
create
the
map.
The
engine
will
initially
be
mapped
with
40
points.
That
data
set
will
be
down
sampled
to
20
points
to
create
torque
and
BSFC
maps.
Interpolation
error
will
be
quantified
by
Page
18
of
51
comparing
the
subset
of
the
40
points
that
were
NOT
used
to
create
the
20­
point
map
with
the
interpolated
value
from
the
20
point
map.

This
task
will
use
the
following
systems:
g)
Two
(
2)
heavy
duty
diesel
engines
(
1
HHDE,
1
LHDE)
for
Subtask
A­
3­
1.
h)
One
(
1)
heavy
duty
diesel
engine
(
1
MHDE)
for
Subtask
A­
3­
2.
i)
Three
(
3)
heavy
duty
diesel
engines
(
1
HHDE,
1
MHDE,
1
LHDE)
for
Subtask
A­
3­
3.
j)
Each
engine
will
be
mapped
at
the
3rd
party
lab
for
torque
and
BSFC
as
a
function
of
speed
and
fuel
commanded,
using
a
test
matrix
of
40
points
distributed
within
the
NTE
zone
The
test
procedure
will
follow
the
guidelines
described
below:

A­
3.1
(
DOE
Testing):
15.
Design
of
Experiments
(
DOE)
matrix:
half
factorial,
resolution
IV,
4
factors,
1
center
point
(
9
pts)
16.
Each
parameter
will
be
evaluated
at
two
conditions
(
min
and
max)
17.
A
center
point
will
be
included
to
investigate
nominal
conditions
18.
Operating
conditions:
five
(
5)
steady
state
engine
conditions
which
are
a
subset
of
the
40­
point
map.
Examples:
A30,
A100,
B50,
C30,
C100
19.
Each
operating
condition:
closed
loop
control
on
engine
"
speed"
and
load
cell
torque.
At
each
condition,
torque,
fuel
flowrate,
and
BSFC
will
be
measured
and
ECM
speed
and
fuel
commanded
will
be
recorded.
20.
Total
number
of
points:
9
x
5
=
45
/
engine
21.
This
task
will
evaluate
error
effects
due
to
intake
air
restriction,
exhaust
gas
restriction,
barometric
pressure,
IMT,
and
EGR
flowrate
A­
3.2
(
Sensitivity
Testing):
Note:
3/
9
 
Now
only
factors
(
5)
and
(
6)
will
be
investigated
through
sensitivity
testing.
Temperature
effects
(
factors
(
1­
4)
will
be
investigated
through
a
single
test.
The
details
below
only
refer
to
(
5)
and
(
6):

a)
This
subtask
will
use
a
simpler
approach
(
not
a
DOE
matrix)
(
6
pts)
b)
Each
parameter
will
be
evaluated
at
three
conditions
(
min,
max,
and
nominal)
c)
Operating
conditions:
three
(
3)
steady
state
engine
conditions:
determined
from
Subtask
A­
3­
1(
subset)
d)
Each
operating
condition:
closed
loop
control
on
engine
"
speed"
and
load
cell
torque.
At
each
condition,
torque,
fuel
flowrate,
and
BSFC
will
be
measured
and
ECM
speed
and
fuel
commanded
will
be
recorded.
e)
Total
number
of
points:
6
x
3
=
18
data
pts/
engine
f)
This
task
will
evaluate
error
effects
due
to
oil
viscosity,
fuel
temperature,
oil
temperature,
coolant
temperature,
humidity,
and
fuel
properties
(
including
cetane
number,
viscosity,
and
API
density).

A­
3.3
(
Linearity
Error
Evaluation):
a)
Determine
40
operating
conditions
within
the
NTE
zone
and
measure
load
cell
torque
and
bsfc
(
using
fuel
flowmeter)
as
a
function
of
engine
speed
and
fuel
commanded.
b)
Downsample
to
20
points
and
use
data
to
create
torque
and
BSFC
maps
c)
Determine
interpolation
error
by
comparing
measurements
of
the
20
points
(
of
the
40)
that
were
NOT
downsampled
to
the
map
that
was
created
from
the
other
20
downsampled
points.
Page
19
of
51
Note
1:
more
detail
is
included
in
Appendix
??.
Note
2:
the
methodology
used
in
this
task
does
not
separate
bias
and
precision
errors.
Both
type
of
errors
will
be
coupled
together.
In
order
to
better
assess
precision
errors,
repeat
testing
would
be
necessary.
However,
it
is
expected
that
the
variability
in
ECM
mapped
torque
and
mapped
BSFC
will
be
greater
than
the
precision
interval
on
a
repeated
test
for
Subtask
A­
3­
1.
It
is
also
expected
that
the
precision
error
(
as
percent
of
point)
in
Subtask
A­
3­
2
will
be
less
than
the
variability
interval
(
as
percent
of
point)
of
Subtask
A­
3­
1.
If
this
does
not
hold,
a
modification
to
the
test
plan
may
be
needed,
which
may
include
repeated
testing.

A­
3.4
Monte
Carlo
Model
The
following
torque
errors
will
be
included
in
the
Monte
Carlo
model
for
torque.
They
will
be
assumed
to
be
independent
and
additive
unless
determined
otherwise.
A
BSFC
model
will
be
constructed
in
a
similar
fashion
based
on
BSFC
data.

A­
3.4.1
Delta
Torque1:
Bias
and
variability
torque
errors
due
to
DOE
parameters
 
Description:
Effect
of
the
following
(
interacting)
parameters
on
mapped
torque:
intake
air
restriction,
exhaust
gas
restriction,
barometric
pressure,
intake
manifold
temperature,
and
EGR
flowrate
 
Data
source:
Subtask
A­
3­
1
 
Data
Plots:
Include
data
from
2
engines,
5
SS
conditions
1)
PEMS
vs.
lab
x­
axis:
%
of
peak
torqueload
cell
(
from
load
cell)
y­
axis:
%
of
peak
torqueECM
(
from
ECM
map)
2)
PEMS
vs.
lab,
subtract
lab
x­
axis:
%
of
peak
torquenom
(
from
load
cell)
y­
axis:
%
peak
torqueECM
­
%
peak
torquenom
3)
Surface
plot:
error
distribution
as
function
of
%
peak
torquenom
x­
axis:
%
of
peak
torquenom
y­
axis:
torque
variability
index,
itorque
from
 
1
to
+
1
z­
axis:
(%
peak
torqueECM
­
%
peak
torquenom)
 
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
iTorque
 
Restrictions
on
input
parameters:
None
 
Error
Calculation:
delta
torque1
=
(%
peak
torqueECM
­
%
peak
torquenom)*
torquepeak
 
PDF
shape
for
delta
torque1:
Uniform
 
Update
frequency
for
delta
torque1:
once
per
NTE
event
(
assumes
all
parameters
are
fairly
constant
during
an
NTE
event)

A­
3.4.2
Delta
Torque2:
Torque
errors
due
to
changes
in
fuel
temperature
NOTE:
3/
9
 
if
kept,
this
will
be
combined
with
other
effects
into
a
delta
torque
2.
Data
analysis
technique
will
change
(
will
use
data
from
engine
warmup
test).

 
Description:
Effect
of
fuel
temperature
on
mapped
torque
 
Data
source:
Subtask
A­
3­
2
 
Data
plots:
Include
data
from
1
engine,
3
SS
points,
3
fuel
temps.:
1)
%
of
point
error
vs.
fuel
temp
Page
20
of
51
For
each
operating
condition:
x­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
(%)
y­
axis:
Tfuel
(
deg.
C)
2)
Surface
plot:
error
distribution
as
a
function
of
%
peak
torque,
Fuel
temp
x­
axis:
:
%
of
peak
torquenom
y­
axis:
Tfuel
yy­
axis:
Tfuel
variability
index,
ifuel
temp
z­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
ifuel
temp
 
Restrictions
on
input
parameters:
Tfuel
<
65
C;
Tracks
with
Tamb?
 
Error
Calculation:
((
TorqueECM
 
Torquenom)/
Torquenom)*
Torquenom
 
PDF
shape
for
delta
torque2:
??
Gaussian
around
a
mean
fuel
temp
??
 
Update
frequency
for
delta
torque2:
once
per
NTE
event
A­
3.4.3
Delta
Torque3:
Torque
errors
due
to
changes
in
oil
temperature
NOTE:
3/
9
 
if
kept,
this
will
be
combined
with
other
effects
into
a
delta
torque
2.
Data
analysis
technique
will
change
(
will
use
data
from
engine
warmup
test).
 
Description:
Effect
of
oil
temperature
on
mapped
torque
 
Data
source:
Subtask
A­
3­
2
 
Data
plots:
Include
data
from
1
engine,
3
SS
points,
3
oil
temps.:
1)
%
of
point
error
vs.
oil
temp
For
each
operating
condition:
x­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
(%)
y­
axis:
Toil
(
deg.
C)
2)
Surface
plot:
error
distribution
as
a
function
of
%
peak
torque,
oil
temp
x­
axis:
:
%
of
peak
torquenom
y­
axis:
Toil
yy­
axis:
Toil
variability
index,
ioil
temp
z­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
ioil
temp
 
Restrictions
on
input
parameters:
Toil
will
likely
be
between
70
and
110
deg.
C;
Does
it
track
with
anything
else?
 
Error
Calculation:
((
TorqueECM
 
Torquenom)/
Torquenom)*
Torquenom
 
PDF
shape
for
delta
torque3:
??
Gaussian
around
a
mean
oil
temp
??
 
Update
frequency
for
delta
torque3:
once
per
NTE
event
A­
3.4.4
Delta
Torque4:
Torque
errors
due
to
changes
in
coolant
temperature
NOTE:
3/
9
 
if
kept,
this
will
be
combined
with
other
effects
into
a
delta
torque
2.
Data
analysis
technique
will
change
(
will
use
data
from
engine
warmup
test).
 
Description:
Torque
errors
due
to
change
in
coolant
temperature
 
Data
source:
Subtask
A­
3­
2
 
Data
plots:
Include
data
from
1
engine,
3
SS
points,
3
coolant
temps:
1)
%
of
point
error
vs.
coolant
temp
Page
21
of
51
For
each
operating
condition:
x­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
(%)
y­
axis:
Tcoolant
(
deg.
C)
2)
Surface
plot:
error
distribution
as
a
function
of
%
peak
torque,
coolant
temp
x­
axis:
:
%
of
peak
torquenom
y­
axis:
Tcoolant
yy­
axis:
Tcoolant
variability
index,
icoolant
temp
z­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
icoolant
temp
 
Restrictions
on
input
parameters:
??
Coolant
temperature
range?
Does
it
track
with
Tamb?
 
Error
Calculation:
((
TorqueECM
 
Torquenom)/
Torquenom)*
Torquenom
 
PDF
shape
for
delta
torque4:
??
Gaussian
around
a
mean?
 
Update
frequency
for
delta
torque4:
once
per
NTE
event
A­
3.4.5
Delta
Torque5:
Torque
errors
due
to
changes
in
humidity
 
Description:
Torque
errors
due
to
changes
in
ambient
humidity
 
Data
source:
Subtask
A­
3­
2
 
Data
plots:
Include
data
from
1
engine,
3
SS
points,
3
intake
air
humidities:
1)
%
of
point
error
vs.
(
intake
air)
humidity
level
For
each
operating
condition:
x­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
(%)
y­
axis:
H
(
grains/
lb
dry
air)
2)
Surface
plot:
error
distribution
as
a
function
of
%
peak
torque,
humidity
x­
axis:
:
%
of
peak
torquenom
y­
axis:
H
yy­
axis:
H
variability
index,
ihumidity
z­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
ihumidity
 
Restrictions
on
input
parameters:
Humidity
ranges?
 
Error
Calculation:
((
TorqueECM
 
Torquenom)/
Torquenom)*
Torquenom
 
PDF
shape
for
delta
torque5:
Uniform
 
Update
frequency
for
delta
torque5:
once
per
NTE
event
A­
3.4.6
Delta
Torque6:
Torque
errors
due
to
changes
in
fuel
properties
 
Description:
Torque
errors
due
to
changes
in
fuel
properties
(
cetane
number,
API
density,
viscosity)
 
Data
source:
Subtask
A­
3­
2
 
Data
plots:
Include
data
from
1
engine,
3
SS
points,
3
fuels:
1)
%
of
point
error
vs.
fuel
(
discrete
points)
For
each
operating
condition:
x­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
(%)
y­
axis:
Fuel
type
number
(
1,
2
or
3)
2)
Surface
plot:
error
distribution
as
a
function
of
%
peak
torque,
fuel
type
number
Page
22
of
51
x­
axis:
:
%
of
peak
torquenom
y­
axis:
Fuel
type
number
yy­
axis:
Fuel
type
variability
index,
ifuel
z­
axis:
(
TorqueECM
 
Torquenom)/
Torquenom
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
ifuel
 
Restrictions
on
input
parameters:
Inputs
can
either
be
fuel
#
1,
fuel
#
2,
or
fuel
#
3
 
Error
Calculation:
((
TorqueECM
 
Torquenom)/
Torquenom)*
Torquenom
 
PDF
shape
for
delta
torque6:
Uniform
 
Update
frequency
for
delta
torque6:
once
per
NTE
event
A­
3.4.7
Delta
Torque7:
Torque
errors
due
to
assumed
linearity
of
torque/
BSFC
map
 
Description:
Access
errors
due
to
linearly
interpolating
the
torque
and
BSFC
maps
 
Data
source:
Task
A­
3­
3
 
Data
plots:
Use
20
points
(
of
the
40
taken)
from
the
mapping
exercise,
that
were
not
used
to
create
the
maps
1)
PEMS
vs.
lab
x­
axis:
%
of
peak
torqueload
cell
(
from
load
cell)
y­
axis:
%
of
peak
torqueECM
(
from
ECM
map)
2)
PEMS
vs.
lab,
subtract
lab
x­
axis:
%
of
peak
torquenom
(
from
load
cell)
y­
axis:
%
peak
torqueECM
­
%
peak
torquenom
3)
Surface
plot:
error
distribution
as
function
of
%
peak
torquenom
x­
axis:
%
of
peak
torquenom
(
Assume
constant
error
distribution
over
torque
range)
y­
axis:
torque
variability
index,
ilinearity
from
 
1
to
+
1
z­
axis:
(%
peak
torqueECM
­
%
peak
torquenom)
(
Again,
this
is
not
a
function
of
%
peak
torque)
 
Model
inputs:
ilinearity
 
Restrictions
on
input
parameters:
None
 
Error
Calculation:
(%
peak
torqueECM
­
%
peak
torquenom)*
torquepeak
 
PDF
shape
for
delta
torque7:
Uniform
 
Update
frequency
for
delta
torque7:
Every
second
A­
3.4.8
Delta
Torque8:
Torque
errors
due
to
non­
deficiency
AECDs,
ECM
software
maps
 
Description:
Torque
errors
due
to
mapping
torque
with
one
set
of
maps
(
fueling,
lug
curve,
etc.),
but
operating
the
engine
on
a
different
map
 
Data
source:
Task
C­
1
 
Data
plots:
See
Rey's
write­
up
 
Model
inputs:
 
Restrictions
on
input
parameters:
 
Error
Calculation:
 
PDF
shape
for
delta
torque8:
 
Update
frequency
for
delta
torque8:
Page
23
of
51
A­
3.4.9
Delta
Torque9:
Torque
errors
due
to
dynamic
measurement
(
instrument
response,
time
alignment)

 
Description:
Variability
and
bias
errors
in
dynamic
torque
measurements.
May
be
added
as
a
BSE
error,
instead
of
a
torque
error.
 
Data
source:
See
Matt's/
Bill's
write­
up
 
Data
plots:
 
Model
inputs:
 
Restrictions
on
input
parameters:
 
Error
Calculation:
 
PDF
shape
for
delta
torque9:
 
Update
frequency
for
delta
torque9:

A­
3.4.10
Delta
Torque10:
Torque
errors
due
to
production
variability
 
Description:
Torque
errors
due
to
using
a
single
map
for
all
in­
use
testing
of
engines
of
the
same
engine
family
and
rating
 
Data
source:
Engine
manufacturer
data
 
Data
plots:
Assuming
there
is
not
a
big
difference
in
production
variabilities
of
different
engine
families
and
ratings
(
ie,
LHDE
and
HHDE
variabilities
are
about
the
same):
1)
%
peak
torque
engine
variation
vs.
nominal
peak
torque
x­
axis:
%
of
peak
torquenom
y­
axis:
%
of
peak
torquemeasured
(
measured
from
different
engines
of
the
same
families
and
ratings)
2)
Variation
vs.
nominal,
subtract
nominal
x­
axis:
%
of
peak
torquenom
y­
axis:
%
peak
torquemeasured
­
%
peak
torquenom
3)
Surface
plot:
error
distribution
as
function
of
%
peak
torquenom
x­
axis:
%
of
peak
torquenom
y­
axis:
torque
variability
index,
iengine
variability
from
 
1
to
+
1
z­
axis:
(%
peak
torquemeasured
­
%
peak
torquenom)
 
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
iengine
variability
 
Restrictions
on
input
parameters:
None
 
Error
Calculation:
delta
torque1
=
(%
peak
torquemeasured
­
%
peak
torquenom)*
torquepeak
 
PDF
shape
for
delta
torque10:
Uniform
 
Update
frequency
for
delta
torque10:
once
per
NTE
event
A­
3.4.11
Delta
Torque11:
Torque
errors
due
deterioration
factor
 
Description:
Torque
errors
due
to
mapping
on
a
new
engine
but
measuring
on
engine
with
XX
mileage
accumulated
 
Data
source:
Engine
manufacturer
data
 
Data
plots:
Include
engine
manufacturer
data:
1)
%
of
point
error
vs.
mileage
x­
axis:
(
Torquewith
mileage
 
Torquenom)/
Torquenom
(%)
y­
axis:
Mileage
2)
Surface
plot:
error
distribution
as
a
function
of
%
peak
torque,
mileage
Page
24
of
51
x­
axis:
:
%
of
peak
torquenom
y­
axis:
Mileage
yy­
axis:
Mileage
variability
index,
imiles
z­
axis:
(
Torquewith
mileage
 
Torquenom)/
Torquenom
 
Model
inputs:
Torquepeak,
%
Peak
Torquenom,
imiles
 
Restrictions
on
input
parameters:
Miles
between
0
and
useful
life
 
Error
Calculation:
((
Torquewith
mileage
 
Torquenom)/
Torquenom)*
Torquenom
 
PDF
shape
for
delta
torque11:
Uniform
 
Update
frequency
for
delta
torque11:
once
every
NTE
event
Page
25
of
51
B­
1
Evaluation
of
the
effect
of
vehicle
shock,
vibration
and
orientation
on
the
performance
of
the
PEMS
B­
1.1
Objective
Evaluate
the
effect
of
vehicle
shock,
vibration
and
orientation
on
the
performance
of
the
PEMS
and
determine
error
factors
for
the
PEMS
due
to
these
effects.

B­
1.2
Background
The
performance
of
the
PEMS
could
be
affected
by
being
in
a
vehicle
which
is
traveling
on
the
roadway
and
is
subject
to
roadway
irregularities
resulting
in
the
transmission
of
shock
and
vibration
to
the
PEMS.
Also
the
location/
orientation
of
the
PEMS
in
the
vehicle
could
also
be
a
factor.

Engine
companies
conduct
sophisticated
tests
to
gather
information
on
the
shocks
and
vibrations
imparted
to
components
in/
on
the
vehicle
as
the
vehicle
traverses
a
variety
of
roadway
terrains.
Miniature
tri­
axial
accelerometers
are
generally
used
to
gather
orthogonal
axis
information
simultaneously.
The
typical
frequency
range
covers
20
to
2000
Hz.
The
minimum
sampling
rate
of
5
kHz
is
usually
used.
The
collected
data
is
reduced
using
an
FFT
(
Fast
Fourier
Transform
Analyzer).
This
data
is
then
used
to
generate
a
shock/
vibration
profile
that
is
used
in
a
vibration
lab
to
simulate
the
impact
of
roadway
shock/
vibration
on
equipment
under
test.
Different
locations
on
the
vehicle
would
generally
have
different
profiles.
It
would
be
impossible
to
have
every
single
location
on
the
vehicle
where
the
PEMS
could
be
located
characterized
for
shock
and
vibration.
Choices
have
to
be
made
from
available
profiles
which
ones
would
be
appropriate
to
use.
Sometimes
new
profiles
need
to
be
generated
to
account
for
special
installations.
This
could
be
an
expensive
proposition.

B­
1.3
Methods
and
Materials
A
test
facility
which
can
provide
the
ability
to
simulate
roadway
shock/
vibration
would
be
needed.
Since
we
want
to
characterize
6
PEMS
units
under
similar
conditions,
it
would
be
beneficial
if
the
test
"
shaker
table"
where
the
PEMS
can
be
mounted
and
subsequently
subjected
to
the
simulated
shock/
vibration
impulses
can
accommodate
them
all
 
if
not,
the
test
would
have
to
be
repeated
a
number
of
times
until
all
PEMS
units
are
characterized.
The
test
would
involve
the
selection
of
a
set
of
appropriate
shock/
vibration
profiles.
Response
accelerometers
need
to
be
installed
near
each
PEMS
mounting
location
and
the
actual
vibration
levels
are
measured
and
compared
against
the
vibration
profile
to
see
if
they
meet
required
tolerances.

Cummins
uses
several
different
vibration
profiles
to
determine
the
effect
of
vibration
on
the
`
Equipment
under
test'.
It
also
has
a
profile
for
shock
test.
The
vibration
profiles
are
plots
of
amplitude
(
in
units
of
G2/
Hz)
vs.
Frequency
(
in
units
of
Hz)
on
a
log­
log
scale.
Break
points
are
plotted
on
the
graph
and
the
profile
is
generated
by
connecting
the
lowest
frequency
break
point
with
the
next
higher
frequency
break
point
until
the
highest
frequency
break
point
is
connected.
The
plots
are
generally
for
baseline
(
unaccelerated)
vibration
levels.
The
unaccelerated
profile
is
the
correct
profile
for
us
to
use
since
we
are
not
trying
to
simulate
product
life
in
a
reasonable
test
time
(
accelerated
time
to
failure).
Different
profiles
are
available
for
equipment
under
test
mounted
in
the
cab
and
on
the
frame.
Current
engineering
specifications
used
at
Cummins
for
reference
show
equipment
mounted
with
different
axis
orientation
but
the
same
profile
is
used
regardless
of
the
axis
orientation.
The
caveat
given
is
that
`
the
random
vibration
profiles
provided,
although
derived
from
actual
heavy
duty
diesel
vehicle
field
data,
are
very
broad
in
scope
Page
26
of
51
and
are
provided
as
a
reference/
starting
point
only'.
Hence
it
would
be
good
to
make
sure
that
the
vibration/
shock
profiles
that
are
used
are
appropriate
for
evaluating
the
performance
of
the
PEMS.

B­
1.4
Quantifying
PEMS
performance
through
the
shock/
vibration
profiles
The
actual
PEMS
performance
evaluation
will
require
the
PEMS
to
be
sampling
zero/
span/
and
a
gas
mixture.
We
could
use
the
same
protocol
that
Matt
Spears
has
suggested
in
his
write
up
for
the
evaluation
of
ambient
pressure
effects,
namely:

Prior
to
executing
this
pressure
sequence,
each
PEMS
will
be
setup
and
stabilized
in
the
room's
first
pressure.
The
setup
will
be
done
according
to
PEMS
manufacturer
instructions,
including
any
warm­
up
time,
zero­
spans
of
the
analyzers
and
the
setup
of
all
accessories
including
flow
meters,
ECM
interpreters,
etc.
Then,
when
the
vibration/
shock
sequence
starts,
the
PEMS'
overflow
sample
ports
will
be
supplied
with
a
sequence
of
gases
from
seven
gas
cylinders.
The
gas
cylinders
purity
and
accuracy
do
not
have
to
meet
1065
Subpart
H
specifications.

Table
3:
Gas
cylinder
contents
Gas
Number
of
AL
size
cylinders1
1.
purified
air
1.03
2.
quad­
blend
span:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
3.
CH4
span,
balance
N2
1.03
4.
NO2
span,
balance
N2
1.03
5.
quad­
blend
audit:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
6.
CH4
audit,
balance
N2
1.03
7.
NO2
audit,
balance
N2
1.03
1AL
size
compressed
gas
cylinders
are
high
pressure
(
2000psi)
and
hold
29.5
liters
of
water.
Considering
the
compressibility
(
Z)
of
certain
gases,
a
safe
approximate
supply
from
one
AL
cylinder
is
4,000
liters
at
atmospheric
conditions.
Assuming
6
PEMS
consuming
10
lpm
each
simultaneously
or
60
lpm,
times
8
hours
equals
28800
liters
per
day
or
7.2
cylinders
per
day.
This
means
that
for
this
experiment
cylinders
might
have
to
be
switched
during
the
test,
while
another
one
of
the
cylinders
is
being
sampled.

N2
is
not
in
the
gas
cylinder
matrix
as
a
zero
quantity
since
N2
would
be
just
like
the
CH4
or
NO2
cylinders
for
the
other
gases,
and
the
quad­
blends
are
just
like
N2
for
CH4
and
NO2.
Gas
cylinder
concentrations
will
be
determined
using
some
of
the
initial
results
of
the
dynamometer
laboratory
PEMS
testing;
scaled
up
or
down
such
that
the
cylinder
span
values
are
near
the
flow­
weighted
average
concentration
at
the
NTE
standards,
and
the
audit
values
are
at
about
half
of
the
span
values.

Each
cylinder
should
be
flowed
long
enough
so
that
at
least
25
stable
readings
are
recorded
for
the
slowest
responding
gas
concentration
output
of
all
the
PEMS.
Note
that
care
should
be
taken
to
position
PEMS
and
to
configure
gas
transport
tubing
to
minimize
transport
delays.
A
reasonable
target
time
to
sample
each
cylinder
should
be
about
1
minute
(
30
seconds
to
stabilize
+
30
seconds
to
record
25
stable
readings),
or
7
Page
27
of
51
minutes
to
cycle
through
all
7
cylinders.
Continuously
repeating
this
7­
minute
cycle
over
an
8­
hr
day
would
result
in
about
68
repeats
per
cylinder
or
about
476
data
points
per
day
per
output
recorded.

The
number
of
vibration/
shock
profiles
selected
would
determine
how
much
testing
would
have
to
be
done
to
get
one
complete
set
of
data
on
all
the
PEMS.
We
then
would
need
to
repeat
the
process
so
that
we
get
at
least
three
complete
sets
of
data.
This
test
should
be
replicated
at
least
3
times,
for
a
total
of
at
least
3
test
days
or
1500
data
points
per
output.
Page
28
of
51
B­
2
Evaluation
of
ambient
temperature
effects
on
PEMS
B­
2.1
Objective
Evaluate
the
effects
of
ambient
temperature
on
PEMS
outputs,
and
establish
error
effects
as
a
function
of
temperature
and/
or
a
rate
of
change
of
ambient
temperature.

B­
2.2
Background
PEMS
are
expected
to
operate
over
a
wide
range
of
changing
ambient
temperatures.
It
is
hypothesized
that
some
of
the
errors
of
the
PEMS
outputs
may
be
a
function
of
changes
in
ambient
temperature.
Therefore,
this
experiment
will
change
the
ambient
temperature
surrounding
PEMS
to
evaluate
its
effects
on
PEMS
outputs.
If
correlation
between
an
output's
error
and
change
in
temperature
exists,
then
the
respective
parameters
in
the
error
model
will
be
modified
to
include
ambient
temperature
change
effects.

B­
2.3
Methods
and
Materials
For
this
experiment
a
well
ventilated
room
capable
of
controlling
a
wide
range
of
temperature
changes
is
needed
(­
5
to
45
°
C).
The
room
must
be
able
to
house
at
least
six
PEMS,
their
power
supplies,
the
PEMS
flow
meters,
cables
and
lines,
plus
seven
different
zero,
audit,
and
span
gas
cylinders,
and
a
gas
switching
system.

The
temperature
changes
will
follow
a
pattern
of
first
soaking
the
PEMS
at
a
constant
room
temperature,
then
ramping
the
room
temperature
to
a
new
temperature,
soaking
the
PEMS
at
that
new
temperature,
and
then
ramping
to
another
temperature.
The
following
sequence
of
temperatures
and
times
were
chosen
to
simulate
a
wide
range
of
real­
world
temperatures
and
changes
in
temperature:

Table
4:
Time
and
temperature
sequence
for
8­
hour
evaluation
of
ambient
temperature
effects
Temperature
Time
Ramp
Rate
Phase
°
C
°
F
min
°
C/
min
Comments
1
Soak
20
68
10
0
Cool
in­
garage
pre­
test
PEMS
operations
2
Ramp
20
­
­
5
68
­
23
5
­
5
Leaving
cool
garage
into
cold
ambient
3
Soak
­
5
23
30
0
Operating
at
cold
temperature
outside
of
vehicle
4
Ramp
­
5
­
10
23
­
50
120
+
0.13
Diurnal
warming
during
cold
day
5
Soak
10
50
40
0
Steady
cool
temperature
during
testing
6
Ramp
10
­
25
50
­
77
5
+
3
Return
to
warm
garage
on
a
cool
day
7
Soak
25
77
40
0
Warm
in­
garage
pre­
post­
test
PEMS
operations
8
Ramp
25
­
45
77
­
113
5
+
4
Leaving
warm
garage
into
hot
ambient
9
Soak
45
113
20
0
Operating
at
hot
temperature
outside
of
vehicle
10
Ramp
45
­
30
113
­
86
120
­
0.13
Diurnal
cooling
during
hot
day
11
Soak
30
86
40
0
Steady
warm
temperature
during
testing
12
Ramp
30
­
20
86
­
68
5
­
2
Return
to
cool
garage
on
a
hot
day
13
Soak
20
68
40
0
Cool
in­
garage
post­
test
PEMS
operations
Page
29
of
51
Temperature­
Time
Profile
for
PEMS
Tesitng
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
40
50
0
1
2
3
4
5
6
7
8
Time
(
hr)
Temperature
(
C)

­
76
­
64
­
52
­
40
­
28
­
16
­
4
8
20
32
44
56
68
80
92
104
116
Temperature
(
F)

_____
Temperature
_
_
_
_
1/
2­
hr
moving
average
dT/
dt
(
C/
hr)
_____
Vertical
gridlines
=
hours
_____
Vertical
gridlines
=
7­
min
gas
cylinder
cycle
times
Figure
1:
Temperature
Time
Profile
for
PEMS
Testing.

Prior
to
executing
this
temperature
sequence,
each
PEMS
will
be
setup
and
stabilized
in
the
room's
first
temperature.
The
setup
will
be
done
according
to
PEMS
manufacturer
instructions,
including
any
warm­
up
time,
zero­
spans
of
the
analyzers
and
the
setup
of
all
accessories
including
flow
meters,
ECM
interpreters,
etc.
Then,
when
the
temperature
sequence
starts,
the
PEMS'
overflow
sample
ports
will
be
supplied
with
a
sequence
of
gases
from
seven
gas
cylinders.
The
gas
cylinders
purity
and
accuracy
do
not
have
to
meet
1065
Subpart
H
specifications
because
PEMS
outputs
will
only
used
for
relative
differences
as
a
function
of
temperature
and
rate
of
temperature
change:

Table
5:
Gas
cylinder
contents
Gas
Number
of
AL
size
cylinders1
1.
purified
air
1.03
2.
quad­
blend
span:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
3.
CH4
span,
balance
N2
1.03
4.
NO2
span,
balance
N2
1.03
Page
30
of
51
5.
quad­
blend
audit:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
6.
CH4
audit,
balance
N2
1.03
7.
NO2
audit,
balance
N2
1.03
1AL
size
compressed
gas
cylinders
are
high
pressure
(
2000psi)
and
hold
29.5
liters
of
water.
Considering
the
compressibility
(
Z)
of
certain
gases,
a
safe
approximate
supply
from
one
AL
cylinder
is
4,000
liters
at
atmospheric
conditions.
Assuming
6
PEMS
consuming
10
lpm
each
simultaneously
or
60
lpm,
times
8
hours
equals
28800
liters
per
day
or
7.2
cylinders
per
day.
This
means
that
for
this
experiment
cylinders
might
have
to
be
switched
during
the
test,
while
another
one
of
the
cylinders
is
being
sampled.

N2
is
not
in
the
gas
cylinder
matrix
as
a
zero
quantity
since
N2
would
be
just
like
the
CH4
or
NO2
cylinders
for
the
other
gases,
and
the
quad­
blends
are
just
like
N2
for
CH4
and
NO2.
Gas
cylinder
concentrations
will
be
determined
using
some
of
the
initial
results
of
the
dynamometer
laboratory
PEMS
testing;
scaled
up
or
down
such
that
the
cylinder
span
values
are
near
the
flow­
weighted
average
concentration
at
the
NTE
standards,
and
the
audit
values
are
at
about
half
of
the
span
values.

Each
cylinder
should
be
flowed
long
enough
so
that
at
least
25
stable
readings
are
recorded
for
the
slowest
responding
gas
concentration
output
of
all
the
PEMS.
Note
that
care
should
be
taken
to
position
PEMS
and
to
configure
gas
transport
tubing
to
minimize
transport
delays.
A
reasonable
target
time
to
sample
each
cylinder
should
be
about
1
minute
(
30
seconds
to
stabilize
+
30
seconds
to
record
25
stable
readings),
or
7
minutes
to
cycle
through
all
7
cylinders.
Continuously
repeating
this
7­
minute
cycle
over
an
8­
hr
day
would
result
in
about
68
repeats
per
cylinder
or
about
476
data
points
per
day
per
output
recorded.

This
test
sequence
should
be
conducted
once.

B­
2.4
Data
Analysis
To
filter
random
noise
from
the
temperature
effects,
the
mean
of
each
25­
recording
segment
will
be
considered
one
data
point.
Subtract
from
each
data
point
its
respective
gas
cylinder
reference
value.
These
are
the
differences
that
will
be
plotted
in
the
Monte
Carlo
surface
plot
as
a
function
of
the
rate
of
change
in
ambient
temperature
and
reference
cylinder
concentration.
For
data
points
that
occur
at
the
same
rate
of
change
in
ambient
temperature
and
concentration,
calculate
a
mean
difference,
and
use
that
mean
difference
in
the
surface
plot.
Page
31
of
51
B­
3
Evaluation
of
ambient
pressure
effects
on
PEMS
B­
3.1
Objective
Evaluate
the
effects
of
ambient
pressure
on
PEMS
outputs,
and
establish
error
effects
as
a
function
of
pressure.

B­
3.2
Background
PEMS
are
expected
to
operate
over
ranges
ambient
pressures.
It
is
hypothesized
that
some
of
the
errors
of
the
PEMS
outputs
may
be
a
function
of
ambient
pressure.
Therefore,
this
experiment
will
change
the
ambient
pressure
surrounding
PEMS
to
evaluate
its
effects
on
PEMS
outputs.
If
correlation
between
an
output's
error
and
pressure
exists,
then
the
respective
parameters
in
the
error
model
will
be
modified
to
include
ambient
pressure
effects.

B­
3.3
Methods
and
Materials
For
this
experiment
a
well
ventilated
room
capable
of
controlling
a
wide
range
of
pressure
changes
is
needed
(
70
to
110
kPa).
The
room
must
be
able
to
house
at
least
six
PEMS,
their
power
supplies,
the
PEMS
flow
meters,
cables
and
lines,
plus
seven
different
zero,
audit,
and
span
gas
cylinders,
and
a
gas
switching
system.

The
pressure
changes
will
follow
a
pattern
of
first
soaking
the
PEMS
at
a
constant
room
pressure,
then
ramping
the
room
pressure
to
a
new
pressure,
soaking
the
PEMS
at
that
new
pressure,
and
then
ramping
to
another
pressure.
The
following
sequence
of
pressures
and
times
were
chosen
to
simulate
a
wide
range
of
real­
world
pressures
and
changes
in
pressure,
which
are
believed
to
be
dominated
by
changes
in
altitude:

Table
6:
Time
and
pressure
sequence
for
8­
hour
evaluation
of
ambient
pressure
effects
Pressure
Time
Ramp
Rate
Phase
kPa
Alt.
ft.
min
ft/
min
Comments
1
Soak
101
89
10
0
Flat
at
sea­
level
2
Ramp
101
­
97
89
­
1203
20
56
Moderate
hill
climb
from
sea
level
3
Soak
97
1203
20
0
Flat
at
moderate
elevation
4
Ramp
97
­
105
1203
­
­
989
60
­
37
Moderate
descent
to
far
below
sea
level
5
Soak
105
­
989
20
0
Flat
at
extreme
low
elevation
6
Ramp
105
­
101
­
989
 
89
20
54
Moderate
hill
climb
to
sea
level
7
Soak
101
89
20
0
Flat
at
sea
level
8
Ramp
101
­
83
89
 
5417
20
266
Extreme
hill
climb
to
near
NTE
limit
9
Soak
83
5417
25
0
Flat
near
NTE
limit
10
Ramp
83
­
75
5417
­
8092
20
134
Rapid
climb
above
NTE
limit
11
Soak
75
8092
20
0
Flat
above
NTE
limit
12
Ramp
75
­
70
8092
­
9883
20
90
Rapid
climb
above
NTE
limit
13
Soak
70
9883
20
0
Flat
far
above
NTE
limit
14
Ramp
70
­
97
9883
 
1203
30
­
289
Extreme
descent
across
NTE
limit
15
Soak
97
1203
20
0
Flat
at
moderate
elevation
16
Ramp
97
­
83
1203
­
5417
15
281
Extreme
hill
climb
to
near
NTE
limit
17
Soak
83
5417
10
0
Flat
at
near
NTE
limit
18
Ramp
83
­
90
5417
­
3244
20
­
109
Rapid
descent
within
upper
half
of
NTE
Page
32
of
51
19
Soak
90
3244
20
0
Flat
within
upper
half
of
NTE
20
Ramp
90
­
97
3244
­
1203
20
­
102
Rapid
descent
to
moderate
elevation
21
Soak
97
1203
20
0
Flat
at
moderate
elevation
22
Ramp
97
­
101
1203
­
89
10
­
111
Rapid
descent
to
sea
level
23
Soak
101
89
20
0
Flat
at
sea
level
Pressure­
Time
Profile
for
PEMS
Tesitng
­
18000
­
16000
­
14000
­
12000
­
10000
­
8000
­
6000
­
4000
­
2000
0
2000
4000
6000
8000
10000
12000
0
1
2
3
4
5
6
7
8
Time
(
hr)
Altitude
(
ft)
&
dA/
dt
(
ft/
hr)
65
75
85
95
105
115
125
135
145
155
Approximate
Pressure
(
kPa)

______
Altitude
(
ft)
_
_
_
_
1/
2­
hr
moving
avg
dA/
dt
(
ft/
hr)
______
Vertical
gridlines
=
hours
______
Vertical
gridlines
=
7­
min
gas
cylinder
cycle
times
Figure
2:
Pressure­
Time
Profile
for
PEMS
Testing
Prior
to
executing
this
pressure
sequence,
each
PEMS
will
be
setup
and
stabilized
in
the
room's
first
pressure.
The
setup
will
be
done
according
to
PEMS
manufacturer
instructions,
including
any
warm­
up
time,
zero­
span­
audits
of
the
analyzers
and
the
setup
of
all
accessories
including
flow
meters,
ECM
interpreters,
etc.
Then,
when
the
pressure
sequence
starts,
the
PEMS'
overflow
sample
ports
will
be
supplied
with
a
sequence
of
gases
from
seven
gas
cylinders.
The
gas
cylinders
purity
and
accuracy
do
not
have
to
meet
1065
Subpart
H
specifications
because
PEMS
outputs
will
only
used
for
relative
differences
as
a
function
of
pressure
and
rate
of
pressure
change:

Table
7:
Gas
cylinder
contents
Gas
Number
of
AL
size
Page
33
of
51
cylinders1
1.
purified
air
1.03
2.
quad­
blend
span:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
3.
CH4
span,
balance
N2
1.03
4.
NO2
span,
balance
N2
1.03
5.
quad­
blend
audit:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
6.
CH4
audit,
balance
N2
1.03
7.
NO2
audit,
balance
N2
1.03
1AL
size
compressed
gas
cylinders
are
high
pressure
(
2000psi)
and
hold
29.5
liters
of
water.
Considering
the
compressibility
(
Z)
of
certain
gases,
a
safe
approximate
supply
from
one
AL
cylinder
is
4,000
liters
at
atmospheric
conditions.
Assuming
6
PEMS
consuming
10
lpm
each
simultaneously
or
60
lpm,
times
8
hours
equals
28800
liters
per
day
or
7.2
cylinders
per
day.
This
means
that
for
this
experiment
cylinders
might
have
to
be
switched
during
the
test,
while
another
one
of
the
cylinders
is
being
sampled.

N2
is
not
in
the
gas
cylinder
matrix
as
a
zero
quantity
since
N2
would
be
just
like
the
CH4
or
NO2
cylinders
for
the
other
gases,
and
the
quad­
blends
are
just
like
N2
for
CH4
and
NO2.
Gas
cylinder
concentrations
will
be
selected
so
that
the
audit
values
are
near
the
flow­
weighted
average
concentration
of
emissions
in
the
raw
exhaust
at
the
NTE
standards,
span
values
will
be
about
twice
the
audit
values,
NO2
will
be
at
half
the
concentration
of
NO,
and
CH4
will
be
at
half
the
concentration
of
C3H8.

Each
cylinder
should
be
flowed
long
enough
so
that
at
least
25
stable
readings
are
recorded
for
the
slowest
responding
gas
concentration
output
of
all
the
PEMS.
Note
that
care
should
be
taken
to
position
PEMS
and
to
configure
gas
transport
tubing
to
minimize
transport
delays.
A
reasonable
target
time
to
sample
each
cylinder
should
be
about
1
minute
(
30
seconds
to
stabilize
+
30
seconds
to
record
25
stable
readings),
or
7
minutes
to
cycle
through
all
7
cylinders.
Continuously
repeating
this
7­
minute
cycle
over
an
8­
hr
day
would
result
in
about
68
repeats
per
cylinder
or
about
476
data
points
per
day
per
output
recorded.

This
test
should
be
performed
once.

B­
3.4
Data
Analysis
To
filter
random
noise
from
the
pressure
effects,
the
mean
of
each
25­
recording
segment
will
be
considered
one
data
point.
Subtract
from
each
data
point
its
respective
gas
cylinder
reference
value.
These
are
the
differences
that
will
be
plotted
in
the
Monte
Carlo
surface
plot
as
a
function
of
ambient
pressure
and
reference
cylinder
concentration.
For
data
points
that
occur
at
the
same
pressure
and
concentration,
calculate
a
mean
difference,
and
use
that
mean
difference
in
the
surface
plot.
Page
34
of
51
B­
4
Evaluation
of
the
effect
of
Electromagnetic
Interference
and
Radio
Frequency
Interference
on
the
performance
of
the
PEMS
B­
4.1
Objective
Evaluate
the
effect
of
Electromagnetic
Interference
(
EMI)
and
Radio
frequency
Interference
(
RFI)
on
the
performance
of
the
PEMS
and
determine
error
factors
for
the
PEMS
due
to
these
effects.

B­
4.2
Background
The
performance
of
the
PEMS
could
be
affected
by
being
in
a
vehicle
which
is
traveling
on
the
roadway
and
is
subject
to
interferences
from
surrounding
EMI/
RFI
signals
 
from
the
vehicle
itself
and
from
items
external
to
the
vehicle.

Electromagnetic/
Radio
frequency
emissions
 
energy
that
is
transferred
through
space
in
the
form
of
electromagnetic/
radio
frequency
waves
resulting
from
the
operation
of
electrical,
radio
and
electronic
equipment.
The
tests
below
refer
to
EMI/
RFI
related
testing.
There
are
other
tests
that
are
carried
out
that
are
electrical
in
nature
such
as:
Over
voltage,
reverse
voltage,
short
circuits,
open
circuits,
power
interrupts,
key
switch
decay
but
will
not
be
considered
here.
These
may
affect
the
PEMS
performance
and
the
`
Steering
Committee'
will
have
to
look
at
the
items
and
determine
if
it
should
also
be
included
in
the
test
plan.

Tests
that
are
usually
used
for
Electromagnetic
compatibility
are:


Radiated
Immunity
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
and
associated
cabling
to
withstand
electric
fields

Radiated
Emissions
 
This
test
method
is
used
to
verify
that
the
electric
field
emissions
from
the
PEMS
and
its
associated
cabling
do
not
exceed
specified
requirements

Conducted
Immunity
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
to
withstand
signals
coupled
onto
input
power
leads

Conducted
Emissions
 
This
test
method
is
used
to
verify
that
electromagnetic
emissions
from
the
PEMS
do
not
exceed
the
specified
requirements
for
power
input
leads,
including
returns

Electrostatic
Discharge
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
to
withstand
electrostatic
discharge
from
the
human
body

Conducted
Transient
Immunity
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
to
withstand
electrical
transients

Electrical
fast
transients
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
and
associated
cabling
to
withstand
short
transients

Surge
Immunity
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
and
associated
cabling
to
withstand
surges
caused
by
switching
and
lightning
transients

Alternator
Noise
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
to
with
stand
transients
where
voltage
differences
are
developed
across
different
current
return
paths
through
the
chassis

Magnetic
field
immunity
 
This
test
method
is
used
to
verify
the
ability
of
the
PEMS
and
associated
cabling
to
withstand
magnetic
fields
resulting
from
nearby
wiring
carrying
high
current
The
list
is
quite
long,
however,
only
a
few
of
these
tests
seem
relevant
to
what
we
need
to
test
for
the
EMI/
RFI
effect.
Also,
based
upon
the
experiences
with
SEMTECH­
D
currently
used
in
in­
use
applications
some
of
the
listed
tests
do
not
seem
to
necessary
since
the
SEMTECH­
D
seems
to
function
reasonably
well
within
the
vehicle.
Radiated
Immunity
seems
to
be
the
main
test
that
needs
to
be
run.
Conducted
Transient
Immunity
and
Electrical
fast
transients
are
other
tests
that
could
also
be
run
based
upon
what
the
`
Steering
Page
35
of
51
Committee'
decides.
The
method
that
is
described
below
is
for
evaluating
the
PEMS
susceptibility
to
Radiated
Immunity.

B­
4.3
Methods
and
Materials
B­
4.3.1
Radiated
Immunity:

Purpose:


The
purpose
of
this
test
is
to
verify
the
ability
of
the
PEMS
and
associated
cabling
to
withstand
electric
fields.
The
test
methodology
would
vary
depending
upon
the
test
site.
Test
Equipment

Typically
the
following
test
equipment
would
be
needed:
Signal
generators,
Power
amplifiers,
Transmit
antennas,
Electric
Field
Sensors,
Measurement
Receiver,
Data
recording
device,
LISNs
(
Line
Impedance
Stabilization
Networks)
and
shielded
enclosure.

For
electric
field
calibrations,
electric
field
sensors
are
required
from
10
kHz
to
1
GHz.

Antennas
should
be
set
up
1
meter
from
the
test
setup
boundary.
The
boundary
includes
all
enclosures
of
the
PEMS
and
the
2
meters
of
exposed
interconnecting
and
power
leads.
For
test
setup
boundaries
greater
than
3
meters,
use
multiple
antenna
positions.

Test
procedure:


Turn
on
the
measurement
equipment
and
the
PEMS
and
allow
sufficient
time
for
stabilization.
It
is
assumed
here
that
the
PEMS
would
be
subject
to
the
sample
gases
per
Matt
Spears
description
and
for
now
the
test
procedure
will
focus
on
the
radiated
immunity
issues.


Set
the
signal
source
to
the
required
modulation
and
using
the
appropriate
amplifier
and
transmit
antenna,
establish
an
electric
field
at
the
test
start
frequency.
Frequency
range
is
from
20
MHz
to
1
GHz.
The
maximum
recommended
field
strength
for
the
PEMS
like
devices
is
30
V/
m.


The
entire
frequency
range
shall
be
scanned.
The
minimum
measurement
time
for
analog
measurement
receivers
during
emissions
testing
shall
be
0.015
sec/
kHz
for
frequencies
in
the
10kHz
to
250
kHz
range,
and
1.5
sec/
MHz
in
the
250
kHz
to
1
GHz
frequency
range.
The
measurement
dwell
time
is
0.015
sec.
The
6dB
bandwidth
is
1
kHz
for
the
lower
frequency
range
and
10
kHz
for
the
upper
frequency
range.
Synthesized
measurement
receivers
shall
step
in
one­
half
bandwidth
increments
or
less.


Monitor
the
PEMS
performance
for
susceptibility
effects.
If
susceptibility
is
noted,
determine
the
level
at
which
the
undesirable
response
is
no
longer
present.


Perform
testing
over
the
required
frequency
range
with
the
transmit
antenna
vertically
polarized.
Repeat
testing
above
20
MHz
with
the
transmit
antenna
horizontally
polarized.
If
multiple
antennas
are
used,
repeat
testing
for
each
transmit
antenna
position.

Data
presentation:


Provide
graphical
or
tabular
data
showing
frequency
ranges
and
field
strength
levels
tested.


Provide
correction
factors
necessary
to
adjust
sensor
output
readings
for
equivalent
peak
detection
of
modulated
waveforms.


Provide
data
listing
any
susceptibility
thresholds
which
were
determined
along
with
their
associated
frequencies.
Here,
the
effect
of
the
radiated
immunity
on
the
zero/
span
and
mixed
gas
concentrations
subjected
to
the
PEMS
(
as
outlined
in
Matt
Spear's
write
up)
will
need
to
be
captured
to
determine
impact
on
PEMS
performance
and
how
the
measurement
error
can
be
modeled.
Page
36
of
51
B­
4.4
Quantifying
PEMS
performance
through
the
EMI/
RFI
evaluation:
The
actual
PEMS
performance
evaluation
will
require
the
PEMS
to
be
sampling
zero/
span/
and
a
gas
mixture.
We
could
use
the
same
protocol
that
Matt
Spears
has
suggested
in
his
write
up
for
the
evaluation
of
ambient
pressure
effects,
namely:

Prior
to
executing
this
evaluation,
each
PEMS
will
be
setup
and
stabilized
in
the
room
environment.
The
setup
will
be
done
according
to
PEMS
manufacturer
instructions,
including
any
warm­
up
time,
zero­
spans
of
the
analyzers
and
the
setup
of
all
accessories
including
flow
meters,
ECM
interpreters,
etc.
Then,
when
the
EMI/
RFI
test
sequence
starts,
the
PEMS'
overflow
sample
ports
will
be
supplied
with
a
sequence
of
gases
from
seven
gas
cylinders.
The
gas
cylinders
purity
and
accuracy
do
not
have
to
meet
1065
Subpart
H
specifications.

Table
8:
Gas
cylinder
contents
Gas
Number
of
AL
size
cylinders1
1.
purified
air
1.03
2.
quad­
blend
span:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
3.
CH4
span,
balance
N2
1.03
4.
NO2
span,
balance
N2
1.03
5.
quad­
blend
audit:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
6.
CH4
audit,
balance
N2
1.03
7.
NO2
audit,
balance
N2
1.03
1AL
size
compressed
gas
cylinders
are
high
pressure
(
2000psi)
and
hold
29.5
liters
of
water.
Considering
the
compressibility
(
Z)
of
certain
gases,
a
safe
approximate
supply
from
one
AL
cylinder
is
4,000
liters
at
atmospheric
conditions.
Assuming
6
PEMS
consuming
10
lpm
each
simultaneously
or
60
lpm,
times
8
hours
equals
28800
liters
per
day
or
7.2
cylinders
per
day.
This
means
that
for
this
experiment
cylinders
might
have
to
be
switched
during
the
test,
while
another
one
of
the
cylinders
is
being
sampled.

N2
is
not
in
the
gas
cylinder
matrix
as
a
zero
quantity
since
N2
would
be
just
like
the
CH4
or
NO2
cylinders
for
the
other
gases,
and
the
quad­
blends
are
just
like
N2
for
CH4
and
NO2.
Gas
cylinder
concentrations
will
be
determined
using
some
of
the
initial
results
of
the
dynamometer
laboratory
PEMS
testing;
scaled
up
or
down
such
that
the
cylinder
span
values
are
near
the
flow­
weighted
average
concentration
at
the
NTE
standards,
and
the
audit
values
are
at
about
half
of
the
span
values.

Each
cylinder
should
be
flowed
long
enough
so
that
at
least
25
stable
readings
are
recorded
for
the
slowest
responding
gas
concentration
output
of
all
the
PEMS.
Note
that
care
should
be
taken
to
position
PEMS
and
to
configure
gas
transport
tubing
to
minimize
transport
delays.
A
reasonable
target
time
to
sample
each
cylinder
should
be
about
1
minute
(
30
seconds
to
stabilize
+
30
seconds
to
record
25
stable
readings),
or
7
minutes
to
cycle
through
all
7
cylinders.
Continuously
repeating
this
7­
minute
cycle
over
an
8­
hr
day
would
result
in
about
68
repeats
per
cylinder
or
about
476
data
points
per
day
per
output
recorded.
Page
37
of
51
We
will
need
to
conduct
the
EMI/
RFI
tests
so
that
we
get
one
complete
set
of
data
on
all
the
PEMS.
We
then
would
need
to
repeat
the
process
so
that
we
get
at
least
three
complete
sets
of
data.

This
test
should
be
replicated
at
least
3
times,
for
a
total
of
at
least
3
test
days
or
1500
data
points
per
output.
Page
38
of
51
B­
5
Evaluation
of
ambient
hydrocarbons
effects
on
PEMS
[???
Loren
Matthews
offered
to
run
this
test
at
the
last
meeting.???]

B­
5.1
Objective
Evaluate
the
effects
of
ambient
hydrocarbons
on
PEMS
FID
outputs,
and
establish
error
effects
as
a
function
of
hydrocarbons.

B­
5.2
Background
PEMS
are
expected
to
operate
over
ranges
of
ambient
hydrocarbons.
It
is
hypothesized
that
zero
error
of
the
PEMS
FID
outputs
may
be
a
function
of
ambient
hydrocarbons.
Therefore,
this
experiment
will
change
the
ambient
hydrocarbons
surrounding
PEMS
to
evaluate
its
effects
on
PEMS
FID
zero
error.
If
correlation
between
a
FID
zero
error
and
hydrocarbons
exists,
then
the
respective
parameters
in
the
error
model
will
be
modified
to
include
ambient
hydrocarbons
effects
on
the
FID
zero
error.

There
are
two
reasons
why
a
FID
might
be
affected
by
ambient
hydrocarbons:
1.
The
FID
uses
ambient
air
as
FID
burner
air.
This
introduces
hydrocarbons
into
the
detector
chamber
from
a
source
other
than
raw
engine
exhaust.
Because
a
FID
uses
burner
air
continuously,
ambient
air
hydrocarbons
from
the
burner
air
will
also
be
present
in
the
reaction
chamber.
2.
The
FID
may
use
ambient
air
as
zero
air
during
over­
the­
road
periodic
zeroing.
This
introduces
into
the
detector
a
second
source
of
ambient
hydrocarbons
during
a
FID
zeroing
procedure
(
in
addition
to
the
burner
air
ambient
hydrocarbons.

Furthermore,
these
effects
become
more
complicated
when
one
considers
that
EPA
regulations
set
HDDE
standards
on
a
non­
methane
hydrocarbons
(
NMHC)
basis.
Because
there
is
no
real­
time
instrument
that
directly
measures
NMHC,
NMHC
is
determined
by
subtracting
a
real­
time
methane
measurement
from
a
real­
time
total
hydrocarbons
(
THC)
measurement.
This
means
two
things:
1.
Ambient
hydrocarbons
will
have
different
effects
on
the
net
results,
depending
upon
what
fraction
of
the
total
ambient
hydrocarbons
is
methane.
2.
A
PEMS
will
have
to
quantify
exhaust
NMHC,
which
by
regulation
can
only
be
done
in
real­
time
by
using
two
FIDs;
one
with
a
non­
methane
hydrocarbon
catalytic
cutter
that
measures
only
methane
(
CH4),
and
one
without
a
cutter
so
that
it
measures
total
hydrocarbons
(
THC).

B­
5.3
Methods
and
Materials
For
this
experiment
a
well
ventilated
temperature­
controlled
room
at
nearly
constant
pressure
is
needed.
The
room
must
be
able
to
house
two
PEMS
(
one
from
each
manufacturer),
their
power
supplies,
the
PEMS
flow
meters,
cables
and
lines,
plus
different
zero,
audit,
and
span
gas
cylinders,
and
two
gas
dividers.

The
burner
air
hydrocarbons
changes
will
follow
a
pattern
of
stabilizing
the
PEMS
FIDs'
burner
air
to
one
of
nine
hydrocarbon
combinations
output
to
an
ambient
pressure
overflow.
After
stabilizing
each
burner
air
hydrocarbon
concentration,
the
zero
will
be
set
for
the
THC
and
CH4
FIDs
using
a
zero
gas
cylinder.
Then
the
PEMS
will
continue
to
quantify
zero
air
from
a
gas
cylinder
at
each
of
ten
(
9,
plus
repeat
of
1st)
different
burner
air
hydrocarbon
combinations.
Record
at
least
25
stable
values
at
each
combination.
Then
switch
to
Page
39
of
51
the
next
of
the
nine
ambient
hydrocarbons
combinations.
Reset
zero
with
the
new
burner
air
hydrocarbons
concentration
overflowing
to
the
burner
air
port.

Repeat
the
entire
zero
quantification
sequence
after
zeroing
with
the
latest
ambient
hydrocarbons
concentration.
Continue
this
series
of
sequences
until
all
combinations
have
been
quantified
and
recorded.

The
series
should
take
about
5
hours
to
complete,
which
should
be
completed
in
one
day:
9
separate
zero
setting
procedures
at
4
minutes
each
(
9
x
4
=~
1/
2
hr)
90
phases
where
hydrocarbons
are
switched
and
zero
air
must
be
stabilized
and
quantified;
3
minutes
each
(
90x3=
4.5
hr).

Table
9:
Hydrocarbons
sequence
for
evaluation
of
ambient
hydrocarbons
effects
Burner
air
hydrocarbons
during
setting
of
zero
Burner
air
hydrocarbons
during
quantification
of
zero
air
Phase
Hexane,
ppm
Methane,
ppm
Hexane,
ppm
Methane,
ppm
1
0
0
2
2
2
3
8
8
4
0
2
5
2
8
6
8
0
7
0
8
8
2
0
9
8
2
10
0
0
0
0
11
2
2
12
8
8
13
0
0
14
2
8
15
8
0
16
0
2
17
2
0
18
8
2
19
0
8
20
2
2
2
2
21
8
8
22
2
2
23
0
0
24
8
2
25
2
0
26
0
8
27
8
0
28
2
8
29
8
8
0
2
Page
40
of
51
30
8
8
31
0
0
32
2
2
33
8
8
34
0
2
35
2
8
36
8
0
37
0
8
38
2
0
39
8
2
40
0
2
0
0
41
2
2
42
8
8
43
0
0
44
2
8
45
8
0
46
0
2
47
2
0
48
8
2
49
0
8
50
2
8
2
2
51
8
8
52
0
0
53
2
2
54
8
0
55
0
2
56
2
8
57
8
2
58
0
8
59
2
0
60
8
0
8
8
61
0
0
62
2
2
63
8
8
64
0
2
65
2
8
66
8
0
67
0
8
68
2
0
69
8
2
70
0
8
0
0
71
2
2
72
8
8
73
0
0
74
2
2
2
8
Page
41
of
51
75
8
0
76
0
2
77
2
0
78
8
2
79
0
8
80
2
2
81
8
8
82
0
0
83
2
2
84
8
0
85
0
2
86
2
8
87
8
2
88
0
8
89
2
0
90
8
0
8
8
Prior
to
executing
this
hydrocarbons
sequence,
each
PEMS
will
be
setup
and
stabilized
at
the
zero
burner
air
hydrocarbons
combination.
The
setup
will
be
done
according
to
PEMS
manufacturer
instructions,
including
any
warm­
up
time,
zero­
spans­
audits
of
the
analyzers
and
the
setup
of
all
accessories
including
flow
meters,
ECM
interpreters,
etc.
Then,
when
the
hydrocarbons
sequence
starts,
the
PEMS'
zero
air
ports
will
be
supplied
with
zero
air
with
the
designated
ambient
hydrocarbons
being
supplied
to
the
burner
air.
The
gas
cylinders
purity,
but
not
accuracy
must
meet
1065
Subpart
H
specifications
because
PEMS
gas
purity
is
important,
but
outputs
will
only
used
for
relative
differences
as
a
function
of
hydrocarbons:

Table
10:
Gas
cylinder
contents
Gas
Number
of
AL
size
cylinders1
1.
purified
air
2.4
2.
C6H14
span,
balance
N2
2.4
3.
CH4
span,
balance
N2
2.4
1AL
size
compressed
gas
cylinders
are
high
pressure
(
2000psi)
and
hold
29.5
liters
of
water.
Considering
the
compressibility
(
Z)
of
certain
gases,
a
safe
approximate
supply
from
one
AL
cylinder
is
4,000
liters
at
atmospheric
conditions.
Assuming
6
PEMS
consuming
10
lpm
each
simultaneously
or
60
lpm,
times
8
hours
equals
28800
liters
per
day
or
7.2
cylinders
per
day.
This
means
that
for
this
experiment
cylinders
might
have
to
be
switched
during
the
test,
while
another
one
of
the
cylinders
is
being
sampled.

Ambient
hydrocarbons
concentrations
will
be
controlled
by
a
gas
divider
to
the
values
in
Table
?,
unless
new
information
about
the
range
of
ambient
hydrocarbons
dictates
a
change
in
the
test
matrix's
values.
Such
information
could
come
from
the
UCR
Ce­
Cert
results
of
Ce­
Cert
trailer
continuous
CVS
background
measurements.
Page
42
of
51
Each
cylinder
should
be
flowed
long
enough
so
that
at
least
25
stable
readings
are
recorded
for
the
slowest
responding
gas
concentration
output
of
all
the
PEMS.
Note
that
care
should
be
taken
to
position
PEMS
and
to
configure
gas
transport
tubing
to
minimize
transport
delays.

This
test
should
be
replicated
only
once.

B­
5.4
Data
Analysis
To
filter
random
noise
from
the
hydrocarbons
effects,
the
mean
of
each
25­
recording
segment
will
be
considered
one
data
point.
In
the
error
model
randomly
sample
from
this
90­
point
data
set
and
use
the
sampled
value
as
a
delta
to
the
NMHC
ppmC
value
in
the
model.
Page
43
of
51
B­
6
Evaluation
of
baseline
repeatability
and
bias
of
PEMS
B­
6.1
Objective
Evaluate
the
baseline
repeatability
and
bias
of
PEMS
with
ambient
conditions
held
constant.

B­
6.2
Background
All
of
the
other
environmental
tests
inherently
incorporate
the
baseline
repeatability
and
bias
of
the
PEMS.
Because
the
Monte
Carlo
simulation
model
adds
all
the
errors
determined
from
the
various
environmental
tests,
it
adds
the
baseline
repeatability
and
bias
of
PEMS
to
the
model
too
many
times.
In
order
to
compensate
for
this
in
the
model,
the
baseline
repeatability
and
bias
of
PEMS
must
be
determined
and
subtracted
from
each
of
the
surface
plots
of
the
environmental
tests.

Note
that
the
baseline
repeatability
and
bias
of
PEMS
is
measured
and
modeled
as
part
of
the
steady­
state
engine
dynamometer
laboratory
experiment.

B­
6.3
Methods
and
Materials
For
this
experiment
a
well
ventilated
EMI/
RFI
shielded
room
capable
of
maintaining
reasonably
constant
temperature
and
pressure
must
be
used.
The
room
must
be
able
to
house
at
least
six
PEMS,
their
power
supplies,
the
PEMS
flow
meters,
cables
and
lines,
plus
seven
different
zero,
audit,
and
span
gas
cylinders,
and
a
gas
switching
system.

Prior
to
executing
this
test,
each
PEMS
will
be
setup
and
stabilized.
The
setup
will
be
done
according
to
PEMS
manufacturer
instructions,
including
any
warm­
up
time,
zero­
spans
of
the
analyzers
and
the
setup
of
all
accessories
including
flow
meters,
ECM
interpreters,
etc.
Then,
the
PEMS'
overflow
sample
ports
will
be
supplied
with
a
sequence
of
gases
from
seven
gas
cylinders.
The
gas
cylinders
purity
and
accuracy
do
not
have
to
meet
1065
Subpart
H
specifications
because
PEMS
outputs
will
only
used
for
relative
differences
as
a
function
of
temperature
and
rate
of
temperature
change:

Table
11:
Gas
cylinder
contents
Gas
Number
of
AL
size
cylinders1
1.
purified
air
1.03
2.
quad­
blend
span:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
3.
CH4
span,
balance
N2
1.03
4.
NO2
span,
balance
N2
1.03
5.
quad­
blend
audit:
CO2,
CO,
NO,
C3H8,
balance
N2
1.03
6.
CH4
audit,
balance
N2
1.03
7.
NO2
audit,
balance
N2
1.03
1AL
size
compressed
gas
cylinders
are
high
pressure
(
2000psi)
and
hold
29.5
liters
of
water.
Considering
the
compressibility
(
Z)
of
certain
gases,
a
safe
approximate
supply
from
one
AL
cylinder
is
4,000
liters
at
atmospheric
conditions.
Assuming
6
PEMS
consuming
10
lpm
each
simultaneously
or
60
lpm,
times
8
hours
equals
28800
liters
per
day
or
7.2
cylinders
per
day.
This
means
that
for
this
experiment
cylinders
might
have
to
be
switched
during
the
Page
44
of
51
test,
while
another
one
of
the
cylinders
is
being
sampled.

N2
is
not
in
the
gas
cylinder
matrix
as
a
zero
quantity
since
N2
would
be
just
like
the
CH4
or
NO2
cylinders
for
the
other
gases,
and
the
quad­
blends
are
just
like
N2
for
CH4
and
NO2.
Gas
cylinder
concentrations
will
be
determined
using
some
of
the
initial
results
of
the
dynamometer
laboratory
PEMS
testing;
scaled
up
or
down
such
that
the
cylinder
span
values
are
near
the
flow­
weighted
average
concentration
at
the
NTE
standards,
and
the
audit
values
are
at
about
half
of
the
span
values.

Each
cylinder
should
be
flowed
long
enough
so
that
at
least
25
stable
readings
are
recorded
for
the
slowest
responding
gas
concentration
output
of
all
the
PEMS.
Note
that
care
should
be
taken
to
position
PEMS
and
to
configure
gas
transport
tubing
to
minimize
transport
delays.
A
reasonable
target
time
to
sample
each
cylinder
should
be
about
1
minute
(
30
seconds
to
stabilize
+
30
seconds
to
record
25
stable
readings),
or
7
minutes
to
cycle
through
all
7
cylinders.
Continuously
repeating
this
7­
minute
cycle
over
an
8­
hr
day
would
result
in
about
68
repeats
per
cylinder
or
about
476
data
points
per
day
per
output
recorded.

This
test
sequence
should
be
conducted
once.

B­
6.4
Data
Analysis
To
filter
random
noise
from
the
temperature
effects,
the
mean
of
each
25­
recording
segment
will
be
considered
one
data
point.
Subtract
from
each
data
point
its
respective
gas
cylinder
reference
value.
Calculate
the
mean
difference
at
each
cylinder
concentration.
This
is
the
baseline
bias.
Subtract
this
value
from
every
data
point
in
every
environmental
test
surface
plot.
This
shifts
all
of
the
surface
plots
to
null
out
any
inherent
PEMS
bias.
This
bias
will
be
modeled
once
as
part
of
the
steady­
state
engine
dynamometer
laboratory
experiment.

Calculate
½
the
span
between
the
5th
and
95th
percentile
differences
at
each
cylinder
concentration.
This
is
the
baseline
repeatability.
Decrease
the
magnitude
of
every
data
point
in
every
environmental
test
surface
plot
by
this
magnitude.
It
is
possible
that
this
adjustment
will
decrease
some
data
point
magnitudes
to
zero.
Keep
any
zero
magnitude
differences
in
the
surface
plots
as
they
indicate
that
the
effect
of
the
environmental
test
was
insignificant
compared
to
the
baseline
repeatability
of
the
PEMS.
The
baseline
repeatability
will
be
modeled
once
as
part
of
the
steady­
state
engine
dynamometer
laboratory
experiment.
Page
45
of
51
C­
1
Error
Assessment
of
ECM­
based
Torque
and
ECM­
based
BSFC,
at
Engine
Manufacturers
individual
laboratories
C­
1.1
Goal
The
objective
of
this
task
is
to
quantify,
at
each
individual
engine
manufacturers
lab,
errors
associated
with
"
estimating"
engine
torque
and
BSFC
through
ECM­
based
parameters.
The
majority
of
these
errors
will
be
estimated
in
task
A­
3,
at
a
3rd
party
engine
lab.
However,
confidential
and
proprietary
parameters
that
may
affect
torque/
bsfc
mapping
will
be
evaluated
separately,
under
this
task
(
C­
1).

Note
1:
All
measurements
uncertainties
associated
with
this
task
are
burden
of
the
engine
manufacturers
only.
If
engine
manufacturers
fail
to
provide
this
data
before
the
end
of
the
environmental
testing
tasks
(
B­
1
through
B­
6),
a
zero
contribution
to
the
bsfc/
torque
error
will
be
assessed
under
this
task.

C­
1.2
Systems
and
Processes
to
be
used:
As
stated
above,
each
engine
manufacturer
will
be
responsible
for
this
task,
and
thus
the
structure
of
testing
and
reporting
will
be
communicated
as
CBI
(
confidential
business
information)
to
the
EPA.

A
list
of
anticipated
parameters
(
not
all
inclusive)
envisioned
to
be
a
part
of
this
task
is
shown
below:
a)
Non­
defficiency
AECD
strategies
that
are
not
captured
by
task
A­
3
b)
Effect
of
multi­
torque
engine
software
on
torque/
bsfc
maps
c)
Effect
of
production
variability
on
torque/
bsfc
d)
Effect
of
engine
deterioration
on
torque/
bsfc
e)
Any
parameters
not
covered
under
3rd
party
lab
testing
C­
1.3
Data
Analysis
EPA
and
CARB
would
consider
information
for
an
additional
allowance
if
error
due
to
AECDs
is
consistent
across
manufacturers.
If
errors
are
infrequent,
large,
or
there
is
a
consistent
bias
in
errors,
EPA
and
CARB
will
expect
those
manufacturers
to
account
for
these
errors
by
creating
more
sophisticated
algorithms
for
calculating
torque/
BSFC
from
ECM
parameters.
Furthermore,
manufacturers
are
invited
to
voluntarily
submit
to
EPA/
CARB
laboratory
information
on
how
non­
deficiency
AECDs
affect
the
accuracy
of
ECMderived
torque/
BSFC
for
NTE
events.

C­
1.4
Deadline
EPA/
CARB
will
not
consider
information
from
this
task
(
C­
1)
if
it
is
submitted
to
EPA/
CARB
after
3
March
2006,
close­
of­
business.
Page
46
of
51
D­
1
Monte
Carlo
Error
Model
and
Measurement
Allowance
D­
1.1
Objective
Use
Monte
Carlo
(
e.
g.
random
sampling)
techniques
in
an
error
model
to
simulate
the
combined
effects
of
all
the
agreed­
upon
sources
of
PEMS
error
incremental
to
lab
error.
Use
the
outputs
of
the
model
to
determine
pollutant­
specific
brake­
specific
additive
measurement
allowances
for
NOx,
NMHC,
and
CO.

D­
1.2
Overview
The
error
model
uses
Monte
Carlo
techniques
to
sample
frequency
distributions
of
independent
variables
(
e.
g.
agreed­
upon
ambient
conditions)
and
other
agreed
upon
frequency
distributions
(
e.
g.
normal,
random
distributions),
which
are
then
used
to
select
error
values
from
the
ranges
of
errors
determined
in
Tasks
A,
B,
and
C.
Note
that
these
error
ranges
will
already
represent
the
error
of
PEMS
incremental
to
the
error
of
the
lab.
After
sampling
the
various
errors,
the
error
model
combines
the
various
PEMS
errors
and
applies
them
to
an
input
data
file,
which
has
a
nominal
set
of
data
needed
to
calculate
many
NTE
events.
This
input
file
is
based
upon
actual
recorded
data.
[???
Matt
Spears
thinks
this
should
be
on­
vehicle
data
from
an
engine
similar
to
the
ones
tested
in
Task
A1
 
maybe
from
Task
E1???].
After
the
errors
are
applied
to
the
input
file,
NTE
results
are
calculated
using
each
of
the
three
agreed­
upon
NTE
calculation
methods.
Then
a
new
set
of
various
errors
are
sampled,
applied
to
the
original
input
data
file,
and
the
NTE
results
are
calculated
again.
This
is
repeated
thousands
or
even
millions
of
times
so
that
the
model
converges
upon
distributions
of
possible
NTE
results
for
each
of
the
original
NTE
events
in
the
input
data
file.
Then
the
95th
percentile
value
is
determined
for
each
NTE
event
distribution
for
each
emission
(
NOx,
NMHC,
and
CO)
and
for
each
calculation
method.
At
this
point
there
are
nine
data
sets
from
three
emissions,
each
calculated
using
three
different
calculation
methods.
Next,
the
NTE
events
are
calculated
by
each
of
the
three
calculation
methods,
but
with
no
error
sampled
or
applied.
These
results
are
considered
the
nominal
values
of
the
NTE
events.
These
nominal
values
are
subtracted
from
each
of
their
respective
95th
percentile
error
values.
The
results
of
this
subtraction
step
is
nine
sets
of
95th
percentile
differences;
for
3
calculation
methods
used
on
each
of
3
emissions.

The
best
calculation
method
is
now
selected
by
pooling
by
calculation
method
all
of
the
differences
of
all
of
the
emissions.
The
one
calculation
method
with
a
mean
value
less
than
the
other
two
methods
is
the
calculation
method
of
choice
for
determining
the
measurement
allowances.

Now
there
are
only
three
data
sets;
one
for
each
emission.
For
each
emission
the
remaining
data
set
of
95th
percentile
(
brake­
specific)
differences
is
plotted
versus
its
nominal
brake­
specific
values
for
the
NTE
events.
A
first­
order
linear
regression
is
performed,
and
if
the
r2<??
and
SEE>??
[???
Bob
Mason,
Bill
Martin???],
then
the
mean
of
the
95th
percentile
brake­
specific
differences
will
be
the
measurement
allowance.
If
the
mean
is
less
than
zero,
then
the
measurement
allowance
will
be
equal
to
zero.
If
the
r2>=??
And
the
SEE
is
<=??
[???
Bob
Mason,
Bill
Martin???],
then
the
regression
line
will
be
used
to
determine
the
measurement
allowance
at
the
following
brake­
specific
NTE
limit
values:
NOx
=
2.0
g/
hp­
hr
NMHC
=
0.21
g/
hp­
hr
CO
=
19.4
g/
hp­
hr
However,
no
extrapolation
of
the
linear
regression
is
allowed.
If
an
NTE
limit
value
is
outside
of
a
data
set
used
to
create
a
linear
regression,
then
the
value
of
the
linear
regression
closest
to
the
NTE
value,
but
within
the
range
of
the
data
set
will
be
used
as
the
measurement
allowance.
Again,
if
this
value
is
less
than
zero,
then
the
measurement
allowance
will
be
equal
to
zero.
Page
47
of
51
D­
1.3
Methods
and
Materials
Microsoft
Excel
combined
with
the
software
add­
in
"@
Risk"
will
be
used
to
create
the
error
model
and
to
calculate
NTE
results
and
the
measurement
allowances.

D­
1.4
Data
Analysis
For
each
of
the
Tasks
in
A,
B,
and
C
an
error
surface
is
created
and
sampled
according
to
agreed­
upon
frequency
distributions.
The
error
surface
represents
an
additive
error
(
or
a
subtractive
error
if
the
sign
is
negative)
to
the
nominal
value.
All
of
the
errors
from
all
of
the
surfaces
are
sampled
and
applied,
and
updating
of
each
sampled
error
occurs
at
a
specified
frequency;
typically
either
once
every
record
(
ie.
second)
or
once
every
NTE
event,
depending
upon
the
nature
of
the
error.

[???
Rey
Agama???
Were
you
planning
on
explaining
or
illustrating
the
generation
of
an
error
surface,
like
in
your
hand­
written
handout???]

[???
Matt
Spears
comment:
I
think
that
once
the
error
model
and
calculation
steps
are
coded
into
Excel,
we
could
enter
some
dummy
error
surfaces,
run
the
model,
plot
some
results,
and
use
the
plots
to
illustrate
this
section???]
Page
48
of
51
E­
1
Validation
of
the
Monte
Carlo
Model
E­
1.1
Objective
Validate
the
Monte
Carlo
model
by
testing
the
PEMS
in
parallel
with
the
CE­
CERT
trailer
and
by
replaying
tests
in
a
laboratory.

E­
1.2
Background
Previous
tests
examined
effects
of
ambient
conditions
on
PEMS
units.
These
effects
have
been
incorporated
into
a
Monte
Carlo
model.
This
test
is
designed
to
verify
the
model
by
comparing
the
in­
use
differences
between
the
PEMS
system
and
the
CE­
CERT
trailer
in
relation
to
model
predicted
differences.
Further
testing
conducted
in
a
laboratory
will
attempt
to
replay
in­
use
testing
by
repeating
engine
conditions
as
closely
as
possible
to
further
validate
the
model.

E­
1.3
Methods
and
Materials
For
this
experiment
two
PEMS
units,
one
Sensors
and
one
Horiba,
will
be
tested
in
parallel
with
the
CECERT
emissions
trailer.
The
trailer
must
be
able
to
measure
regulated
emissions
and
CO2,
ambient
hydrocarbons
(
methane
and
NMHC),
humidity,
temperature,
and
pressure
and
have
adequate
data
acquisition
capabilities
to
capture
the
additional
measurements
discussed
herein.
The
trailer
will
be
tested
at
SwRI
to
insure
that
a
reasonable
level
of
confidence
can
be
placed
on
the
reported
NTE
results.
These
tests
will
include
a
full
1065
audit
and
parallel
or
back­
to­
back
testing
with
a
1065
compliant
laboratory
at
SwRI
over
the
"
NTE
cycle"
described
in
"
Evaluation
of
PEMS
Repeatability
Error
Due
to
Dynamic
Response,"
Task
"?­#."
In
addition,
the
trailer
will
be
validated
over
the
conditions
present
during
the
over­
the­
road
test
by
measuring
zero,
span,
and
audit
gases
while
traveling
over
the
designated
routes.

The
PEMS
will
be
installed
and
calibrated
as
specified
by
the
manufacturer
in
a
Class
8
truck.
Temperature
sensors
will
be
installed
inside
the
protective
enclosure
housing
PEMS
analyzers,
near
exhaust
flow
meter,
and
in
the
ambient
air
stream.
A
Rohde
&
Schwarz
Spectrum
Analyzer
FSH3
or
similar
unit
will
measure
electromagnetic
spectrum
and
power.
Two
3­
axis
vibration/
shock
transducers
will
measure
vibration/
shock
for
the
PEMS
analyzer
unit
and
the
flow
tube.
The
accelerometers
and
the
spectrum
analyzer
should
have
previously
been
used
during
Phase
B­
1
and
B­
4
of
the
environmental
testing.

Test
routes
will
be
designed
to
meet
the
limits
reasonable
expected
to
be
found
in
use
or
to
exceed
the
NTE
limits.
The
PEMS
will
be
mounted
in
two
locations,
inside
the
cab
and
behind
the
cab,
in
order
to
maximize
environmental
differences.
The
vehicle
will
be
driven
over
3
routes,
each
design
to
test
particular
limits
that
are
expected
to
effect
results
during
the
environmental
tests
B­
1
to
B­
7.
 
Route
1­­
Route
starts
in
the
morning
in
room
temperature
garage.
Vehicle
is
driven
into
cold
ambient
conditions
of
less
than
32
deg
F.
Vehicle
is
operated
throughout
day
in
a
warm
location
where
temperatures
exceed
100
deg
F.
Vehicle
returns
to
cooler
ambient
temperatures.
 
Route
2­­
Vehicle
travels
from
sea
level
to
a
high
altitude
exceeding
6000
ft
and
returns
to
sea
level.
 
Route
3­­
Vehicle
is
operated
in
locations
were
following
conditions
are
known
to
exist:
high
ambient
HC,
high
EMI/
RFI,
rough
road
surface.

Several
weeks
worth
of
testing
are
necessary
to
validate
the
Monte
Carlo
Model.
The
1065
audit
of
the
CECERT
trailer
may
take
several
weeks
in
itself.
The
NTE
cycle
test
will
consist
of
a
20­
minute
cycle
repeated
20
times
on
one
engine,
a
total
of
approximately
2
test
days.
For
the
on­
road
portion
of
the
testing,
Page
49
of
51
it
is
likely
that
Route
1
will
take
a
complete
test
day
and
that
Routes
2
and
Route
3
can
be
completed
in
one
test
day.
This
requires
2
test
days
per
PEMS,
per
mounting
location,
plus
2
days
of
on­
road
CE­
CERT
validation
testing,
for
a
total
of
10
test
days
of
on­
road
CE­
CERT
trailer
operation.
If
more
then
one
PEMS
is
available,
several
systems
could
be
tested
in
parallel.
While
two
flow
meters
in
the
exhaust
could
affect
results,
it
is
likely
that
this
measurement
already
experiences
reduced
ambient
effects
because
the
CE­
CERT
trailer
captures
the
exhaust
and
that
any
further
deviation
from
the
true
operating
conditions
is
immaterial.
Also,
two
exhaust
mass
flow
meters
are
not
necessary
if
one
is
attempting
to
account
for
analyzer
location
or
orientation.
Such
parallel
testing
could
potentially
reduce
the
number
of
test
days,
or
increase
the
amount
of
data
collected
for
all
or
parts
of
the
model.

Additional
testing
will
be
conducted
in
a
laboratory,
where
selected
tests
will
be
"
replayed,"
while
the
PEMS
is
maintained
at
laboratory
conditions.
The
engine
will
be
operated
as
close
as
possible
to
previously
recorded
tests
from
the
on­
road
portion
of
the
validation
testing.
Controlled
conditions
will
include
engine
speed/
load,
ambient
conditions,
and
any
other
condition
that
can
be
repeated
with
a
reasonable
level
confidence.
For
each
PEMS,
two
test
days
will
be
replayed,
for
a
total
of
4
test
days.

Table
12:
Monte
Carlo
Model
Validation
CE­
CERT
Validation
Test
Day
1065
Audit
xx
SwRI
CE­
CERT
Validation
NTE
Cycle
xx
Route
1
xx+
1
On­
Road
CECERT
Validation
Route
2,
Route
3
xx+
2
On­
Road
CE­
CERT
Tests
Test
Day
Sensors/
Route
1
xx+
3
Behind
Cab
Route
2,
Route
3
xx+
4
Sensors/
Route
1
xx+
5
In
Cab
Route
2,
Route
3
xx+
6
Horiba/
Route
1
xx+
7
Behind
Cab
Route
2,
Route
3
xx+
8
Horiba/
Route
1
xx+
9
In
Cab
Route
2,
Route
3
xx+
10
Laboratory
Replay
Test
Day
Simulate
Route
1
xx+
11
Sensors
Simulate
Routes
2
&
3
xx+
12
Simulate
Route
1
xx+
13
Horiba
Simulate
Routes
2
&
3
xx+
14
E­
1.4
Data
Analysis
The
difference
between
the
PEMS
results
and
the
CE­
CERT
trailer
results
will
be
compared
to
the
error
predicted
by
the
Monte
Carlo
model.
Data
on
ambient
conditions
must
be
analyzed
to
insure
that
the
model
was
fully
exercised.
Special
consideration
will
be
paid
to
conditions
where
the
PEMS
was
found
to
be
sensitive
to
its
environment.
If
it
was
found
that
the
PEMS
did
not
see
conditions
that
are
likely
to
occur
in
the
field
that
are
known
to
increase
erroneous
reporting,
then
more
testing
is
required.
Page
50
of
51
To
validate
the
Monte
Carlo
Model
data
must
be
run
through
the
model
and
the
model's
results
must
predict
the
actual
test
results.
To
run
data
forwards
through
model,
raw
concentration
must
be
known.
This
requires
the
CE­
CERT
trailer
to
either
measure
raw
exhaust
concentration
or
determine
dilution
ratio
accurately
enough
to
calculate
the
raw
concentration.
If
concentration
can
be
established
with
a
reasonable
level
of
confidence,
then
the
model
can
be
validated
as
follows
by
comparing
data
in
brake
specific
units
(
using
the
work
recorded
by
the
PEMS
for
both
the
CE­
CERT
and
PEMS
data)
or
in
fuel
specific
units.
Raw
exhaust
concentration,
ambient
conditions,
and
exhaust
flow
are
fed
to
the
model.
Since
ambient
effects
are
incorporated
as
a
distribution
of
error,
the
model
must
be
run
many
times
for
each
NTE
event.
The
model
then
produces
a
likely
distribution
of
error
expected
from
the
PEMS.
If
95%
(
or
maybe
90%­
99%)
of
the
PEMS
recorded
NTE
events
fit
within
the
model
predicted
NTE
distribution
then
the
model
can
be
considered
validated.
Data
from
20
pseudo
NTE
events
are
plotted
below
in
the
graph
titled
"
Brake
Specific
NTE
Events."
The
"
model"
predicted
a
likely
error
distribution
from
the
CE­
CERT
measured
data
and
ambient
conditions.
The
PEMS
data
is
compared
to
the
error
distribution,
and
if
enough
NTE
events
are
within
the
range
the
model
is
deemed
validated.

Brake
Specific
NTE
Events
1.0
1.5
2.0
2.5
3.0
3.5
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
NTE
Event
Brake
Specific
Emission
CE­
CERT
Measured
PEMS
Measured
Model
predicted
95%
confidence
limits
Figure
3:
Model
Validation
NTE
Events
If
it
is
determined
that
the
CE­
CERT
trailer
is
unable
to
accurately
determine
raw
exhaust
concentration
then
the
model
cannot
be
run
forward.
Instead
the
PEMS
recorded
data
can
be
run
through
the
model
as
if
it
were
absolutely
correct.
Then
the
model
will
predict
a
distribution
of
error
that
is
likely
to
occur.
This
error
distribution
should
overlap
with
CE­
CERT
recorded
data
as
long
as
the
error
is
random.
If
there
is
a
bias,
then
the
PEMS
will
record
the
bias
and
the
model
will
predict
a
further
bias
in
the
same
direction,
misrepresenting
the
comparison
 
not
ideal.
Page
51
of
51
One
could
also
compare
the
NTE
events
recorded
by
the
PEMS
and
the
CE­
CERT
trailer.
Below
is
a
graph
that
shows
PEMS
NTE
events
minus
the
Accuracy
Margin
versus
the
CE­
CERT
trailer
results.
If
95%
(
or
maybe
90%­
99%)
of
the
PEMS
values
are
less
than
the
CE­
CERT
NTE
events,
then
the
margin
seems
to
be
correct.
Unfortunately
this
does
not
incorporate
the
model
directly,
but
it
gives
evidence
that
the
model's
end
result,
the
Accuracy
Margin,
is
suitable.
This
test
fails
to
test
the
intricacies
of
the
model
and
may
be
too
broad
for
comfort.

Brake
Specific
NTE
Emissions
1.0
1.5
2.0
2.5
3.0
3.5
1.0
1.5
2.0
2.5
3.0
3.5
CE­
CERT
NTE
BSE
Data
PEMS
NTE
BSE
Data
­
Measurement
Margin
Figure
4:
PEMS
vs
Ce­
Cert
Brake­
Specific
Data.
