Bruce Anderson

and the

AAFEX Science Team

Sampling System Effects on Emission Measurements: Airing our Dirty

Laundry

AAFEX Data Workshop—Orlando

January 8, 2009

Sampling lines:

• What are the size dependent particle losses in the sampling lines?

• Do the particle losses change as the lines age?

• Are gas phase measurements influenced by the sampling lines?

• Do volatile particles nucleate and grow in the sample lines?

Sampling probes:

• Are particles lost in the 1-m probes due to inertial impaction or thermophoresis?

• Are the complex “particle” probes required for representative particle sampling or are existing “gas” probes good enough

• Are particle size distributions skewed by non-isokinetic sampling

• Does chemistry occur on the hot probe surfaces that destroys soot and carbon species

Sampling System Questions addressed during AAFEX

Particle Loss to Sampling Lines a Major Concern

Common aerosol sampling lines varied from 24 to 46 m long

Right 1 M LineLine Line  Tube  Residence (s) Residence (s)

Diameter (in) Length (in) Volume (l) 75 lpm 125 lpm0.04 0.02 0.00 0.00 0.000.25 1.55 0.05 0.04 0.020.31 1.84 0.09 0.07 0.040.56 0.03 0.00 0.00 0.000.31 0.54 0.03 0.02 0.010.62 19.92 4.62 3.70 2.22

24 meters 4.8 liters 3.8 seconds 2.3 seconds

Left 1 M LineLine Line  Tube  Residence (s) Residence (s)

Diameter (in) Length (m) Volume (l) 75 lpm 125 lpm0.04 0.02 0.00 0.00 0.000.25 1.53 0.05 0.04 0.020.31 1.94 0.09 0.07 0.040.56 0.03 0.00 0.00 0.000.31 1.33 0.06 0.05 0.030.62 35.45 8.23 6.58 3.95

40 meters 8.4 liters 6.75 4.05

Left 30 M LineLine Line  Tube  Residence (s) Residence (s)

Diameter (in) Length (in) Volume (l) 75 lpm 125 lpm0.62 46 9.03 7.23 4.34

46 meters 9.0 liters 7.2 seconds 4.3 seconds

Right 30 M LineLine Line  Tube  Residence (s) Residence (s)

Diameter (in) Length (m) Volume (l) 75 lpm 125 lpm0.62 38.2 7.44 5.95 3.57

38.2 meters 7.4 liters 6 seconds 3.6 seconds

Re > 32000 in all lines—Flows fully turbulentMeasured residence times 3x greater than calculated

Aerosol Sampling Line Characteristics

MST

ARI

NASA

3/8” St. St. Manual Valves

HEPA1/2 “ Air Actuated Ball Valves

Manifold

pump

Critical Flow Orifice

Right rake

Background

Left rake

30 m Right

30 m Left

Valve Box

3/8” Electrically Actuated Solenoids

AFRL

Vent

Sample Manifold Located in MST Trailer

Valve box plumbed to draw exhaust sample continuously though transmission lines to avoid stagnation; this could have led to soot build-up on surfaces

1 m left

30 m left

1 m right

30 m right

Calculations Suggest Lines have Small Impact

UTRC model assumes clean, smoothbore tubing; Predicts >90% mass transmission

LongDMA

ballast

Furnace

Contains Crucible w/NaCl

Pump

20 lpm

Filter

200 lpmTSI FM

hepa Eductor

N2 gas

regulator

hepa

Sampling LineTSI 3772

P

TSI 3772

P

RotameterMetering Valve (MV)

Line Losses Measured at End of Mission

Monodisperse particles generated and injected into sample lines, measured at each end using identical particle counters

10 1000.0

0.2

0.4

0.6

0.8

1.0

50%

30 m Left 30 m Right

Sam

ple

Line

Tra

nspo

rt Ef

ficie

ncy

Diameter (nm)

10 1000.0

0.2

0.4

0.6

0.8

1.0

1 m Left 1 m Right

Sam

ple

Line

Tra

nspo

rt Ef

ficie

ncy

Diameter (nm)

50%

Transmission Efficiencies Lower than Predicted

Lines from 30 m Inlet Probes

Lines from 1 m sampling rakes

Measured Particle Transmissions range from 50 to 90%

Clean Lines Typically Exhibit Fairly Low Losses

AAFEX Line Losses Assessed after 35 hours of Engine Runs

Penetration Efficiency of APEX‐3 Sample Lines

10 1000.0

0.2

0.4

0.6

0.8

1.0

30 m 1 m

Sam

ple

Line

Tra

nspo

rt E

ffici

ency

Diameter (nm)

50%

Results suggest lines should be frequently cleaned or that back-purging lines might be better than pulling continuous flow

Evidence Suggests Line Loses Increased with Age

0 1 2 3 4 5 6 70.0

0.2

0.4

0.6

0.8

1.0

R

elat

ive

Bla

ck C

arbo

n E

I

Experiment Day

Slope = -5%/day

#2 engine BC emissions at 30 m at 65 and 85% power

Lines also Influence Gas Measurements

Sticky molecules take time to come to equilibrium with tubing walls

Lag in signal depends on length and age of tubing

Decreasing vapor pressure

Time Constants Derived from Decay Curves

Low Volatility Gases Best Measured with Short Sampling Lines

10 20 30 40 50 60 70 800

100

200

300

400

500

Mas

s Em

issi

on In

dex

(mg/

kg)

Engine Power (%)

10 20 30 40 50 60 70 801015

1016

1017

Aero

sol N

umbe

r EI (

#/kg

)

Engine Power (%)

30 m Probe

30 m Probe

1 m Probe

1 m Probe

Do Lines Promote Volatile Aerosol Formation and Growth?

Number EIs typically 10 to 20 times higher in 30 m samples than 1 m due to nucleation of volatile aerosols

#2 engine, JP-8 Fuel

#2 engine, JP-8 Fuel

Volatile aerosol mass sometimes much greater than BC mass in 30 m samples

All near-field observations to date made using long sampling lines

Line Effects Assessed by Comparing Box vs. Truck

Sampling Line Residence Times <1 sec vs. 16 seconds

~1 m

29.5 m

0 20 40 60 80 1000.0

2.0x1016

4.0x1016

6.0x1016

8.0x1016

1.0x1017

1.2x1017

Deathbox EEPS NASA EEPS

Num

ber E

I (#/

kg)

Engine Power

Jan 28, morning, JP-8 fuel

0 20 40 60 80 1000

50

100

150

200

250

300

350

Deathbox EEPS NASA EEPS

Volu

me

EI (u

m3/

kg)

Engine Power

Jan 28, morning, JP-8 fuel

Measurements Indicate Volatile Aerosols Nucleate in Plume

Volume EIs Comparable

Deathbox Number EIs 2X higher than Truck Values

Sampling Line Residence Times <1 sec for box, 16 seconds for truck

10 100

0

100

200

300

400

500

600

UTRC NASA

V

olum

e E

I (um

3/kg

)

D iameter (nm)

Jan 28, 7% power

10 100

0

20

40

60

80

100

120

140

UTRC NASA

Vol

ume

EI (

um3/

kg)

D iameter (nm)

Jan 28, 45% power

10 100

0

20

40

60

80

100

120

UTRC NASA

Vol

ume

EI (

um3/

kg)

D iameter (nm)

Jan 28, 65% power

10 100-20

0

20

40

60

80

100

120

140

160 UTRC NASA

Vol

ume

EI (u

m3/

kg)

D iameter (nm)

Jan 28, 85% power

Sampling Lines Influence Shape of Size Distributions

Small aerosols are lost to tubing surfaces or undergo coagulation

Does it matter if probe velocity matches the exhaust velocity?

Calculations suggest mismatched velocities may effect large particle numbers

Number and mass EIs show a decreasing trend 

with sample concentration (except for jump at highest point)

Isokinetic Sampling Effects Assessed (imperfectly) by Varying Dilution Flow

Data recorded at 85% thrust where exhaust velocity -> mach 1

Increased dilution flow raises pressure in probe, reduces inlet velocity; iso‐kinetic effects convolved with changing line losses 

Number EI

Mass EI

~5X change in flow velocity

0 50 100 150 200 250 300 350

0.0

0.5

1.0

1.5

2.0

2.5

3.0

CO2-1100 CO2-1900 CO2-2900 CO2-3700 CO2-530

Tota

l Aer

osol

Vol

ume

EI (

um3/

kg)

Diameter (nm)

10 100

0.5

1.0

1.5

2.0

2.5

3.0

3.5

530 ppm 1100 ppm 1900 ppm 2900 ppm

Rat

io o

f Vol

ume

Dis

tribu

tions

(SM

PS

1)

Diameter (nm)

85% Power, Ratio of X size distribution to 3700 ppm size dist

Results are inconclusive—further study warranted; better experiment needed

Large Particles Enriched, but is Flow Velocity the Cause?

Total sample flow changes by up to 75% across dilution range; this may be primary cause for changes in number and mass EI

Mean Diameter

AAFEX WORKSHOPOrlando FloridaBruce Anderson, NASA LaRC Jan 8, 2010

Are Particles Lost to the 1 m inlet probes?

Probe Losses Assessed by Comparing 1 and 30 m NV Particle EIs

Recall that 50% of particles are lost in 30 m sample line, only 10% in 1 m line

Mass EI

Number EI

Left Engine--All DataRed=Individual Points

Blue=average

30 m EIn values high at low power due to re-condensation

of volatile material after heater; no evidence of this at

>45% power

Right Engine--All DataRed=Individual Points

Blue=Average

30 m EIn values high at low power due to re-condensation

of volatile material after heater; no evidence of this at

>45% power

Losses More Pronounced on Right Engine because of Similar Sampling Line Transmission Characteristics

Losses may be caused by impaction, diffusion, thermophoresis or catalytic oxidation

Mass EI

Number EI

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0-2 0

0

2 0

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

1 6 0

1 8 0

2 0 0

3 0 m 1 m

Non

vola

tile

Vol

ume

EI (

um3/

kg)

D ia m e te r (n m )

L e f t E n g in e , 8 5 % P o w e r , J 2 7

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0

0

5 0

1 0 0

1 5 0

2 0 0

3 0 m 1 m

Non

vola

tile

Vol

ume

EI (

um3/

kg)

D ia m e te r (n m )

L e f t E n g in e , 8 5 % P o w e r , J 2 8

0 5 0 1 0 0 1 5 0 2 0 0-2 0

0

2 0

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

1 6 0

1 8 0

3 0 m 1 m

Non

vola

tile

Vol

ume

EI (

um3/

kg)

D ia m e te r (n m )

L e f t E n g in e , 8 5 % P o w e r , J 2 9

0 5 0 1 0 0 1 5 0 2 0 0 2 5 0-2 0

0

2 0

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

1 6 0

1 8 0

3 0 m 1 m

Non

vola

tile

Vol

ume

EI (

um3/

kg)

D ia m e te r (n m )

L e f t E n g in e , 8 5 % P o w e r , J 3 0

1 m values at sizes >100 nm always lower; is this caused by inertial losses at 1 m or by contributions from background aerosols at 30 m?

Loss of Large Particles at 1 m Drive Lower Mass EIs

To examine benefit of dilution probe, a gas probe was re‐plumbed to introduce 

dilution gas ~1 m downstream of the inlet tip, then data was recorded as we 

switched back and forth between the aero and gas probes at different power settings

Is it necessary to use “dilution” probes for particle sampling?

Special rakes required to support probes, add greatly to cost and complexity of tests

= Water

= Diluent

A

A

Section A-A= Sample

AEDC Dilution Probe

Right Engine‐‐JP‐8 FuelRed=Aerosol ProbeBlue=Gas Probe

Number and Mass EIs Roughly Equivalent

Right EngineJP‐8 Fuel

Red=Aerosol ProbeBlue=Gas Probe

Some Indication that Nucleation Occurs in Gas Probe

Line losses greater than predicted, probably due to build-up of surface contaminants during testing. Losses clearly a major problem—sampling approaches need to be strictly defined.

Sticky gases require a long time to reach equilibrium with sample line walls, makes accurate determination of alcohols, acids and PAHs difficult.

Volatile aerosols condense in plume, but continue to form and grow within the long sampling lines.

Evidence suggests that probe inlet velocity matters, but is difficult to resolve from other sampling effects.

Results clearly indicate that particles are lost to the 1 m inlet probes, but the exact mechanisms responsible are difficult to define.

Gas probes yield OK particle mass emissions data, but can lead to number EI enhancements due to nucleation in undiluted section

Summary