nist’swind tunnel and stack simulator
TRANSCRIPT
![Page 1: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/1.jpg)
Scale-Model Smokestack Simulator
Probe Calibrations and RATA Results from
NIST’s Wind Tunnel and Stack Simulator
NIST WIND TUNNEL
Scale-Model Smokestack Simulator
NIST Workshop Improving Measurement for Smokestack Emissions
June 28 and 29, 2017
Gaithersburg, MD
Iosif Shinder and Aaron Johnson
![Page 2: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/2.jpg)
Goals… To Answer the Following Questions
1. What is the accuracy of S-probe RATA?
2. Can 3-D probes make the flow RATA more
accurate?
• By how much?
• What parameters need to be considered in a 3-D probe
calibration?
• How accurate are non-nulling methods?
![Page 3: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/3.jpg)
Goals… To Answer the Following Questions
1. What is the accuracy of S-probe RATA?
overpredicts by 5% - 10% Depending on Pitch
2. Can 3-D probes make the flow RATA more
accurate? Yes!
• By how much? Expected accuracies of 1% - 3%
• What parameters need to be considered in a 3-D probe
calibration? Pitch, Yaw, Reynolds number & Turbulence
• How accurate are non-nulling methods? 1% to 3% for
± 12° Pitch and ± 33° Yaw
![Page 4: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/4.jpg)
What is the Acceptable Accuracy of
Stack Flow Measurements?
1) Accuracy 10 %
o relax and skip this presentation ☺.
2) Accuracy 5 %
o S-probes are not sufficient;
o 3-D probes can provide better accuracy
o better continue pay attention
3) Accuracy 1-2 %
o challenging … but we get there if NIST and Industry cooperate
o Two new parameters must be incorporated in probe calibration
❖ Reynolds number (Re)
❖ Turbulence Intensity (Tu)
![Page 5: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/5.jpg)
NIST Wind Tunnel
• Closed loop recirculating wind tunnel
• Test volume: 6.6 ft long 4.9 ft wide 3.9 ft high
– Large test volume small wall effects
• Uncertainty = 0.42% for airspeeds from 16 – 100 ft/s (5 – 30 m/s)
– Uniform flow along tunnel axis (1-dimensional flow)
– Automated staging to control pitch and yaw angles of pitot probes
– Calibrations are automated
![Page 6: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/6.jpg)
Prism Probe
P4
P3
P5
P2
P1
S-probe
D = 1inch
Spherical Probe
P1
P2
P5
P4
P3
Calibration of 3 Conventional Probes
• Diameter of each probe shaft is D = 1 inch
• Length of each probe shaft is 6 ft
![Page 7: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/7.jpg)
Probe Installation in Wind Tunnel
Prism
Probe
LDA
Sensing
Volume
Yaw Range ± 45°
LDA
Beams P4
P3
P5
P2
P1
Feed Through
to Automated
Traverse which
sets pitch and yaw
Pitch = 0 °
![Page 8: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/8.jpg)
NIST Wind Tunnel 4 Axis Traverse System
Yaw Angle (b )
bFeed Through
Probe
• Installed on the side of Wind Tunnel
• Single axis rotation sets Yaw Angle (b)
• 2 Linear motions and a rotation adjust pitch angle (a ) while
maintaining probe head at same position in Wind Tunnel
• Completely automated and synchronized with airspeed measurement
software
![Page 9: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/9.jpg)
Probe Installation Inside of Wind Tunnel
Prism
Probe
LDA
Sensing
Volume
LDA
Beams
![Page 10: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/10.jpg)
S-probe Calibration
pC
b
std cos( )
PV pC
During
Calibration
(b = 0)
0.90
0.88
0.86
0.84
0.82
0.80
p,default 0.84C
0 20 80 100
a = 5°
a = 10°
a = 0°
a = -5°
a = -10°
, [ft / s]V40 60
• S-probe calibration coefficient depends on velocity
• S-probe calibration coefficient depends on pitch (often
neglected in S-probe calibrations)
![Page 11: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/11.jpg)
Calibration Method for Prism (or Spherical) Probe (EPA Method 2F)
1) Set airspeed; VNIST =16 to 100 ft/s (16.5 ft/s steps)
2) Set probe pitch angles: a = -45° to 45° (3° steps)
3) Rotate probe until P2 = P3 to determine Yaw Angle (b )
P4
P3
P5
P2
P1
Prism 4) Measure Pitch Calibration Factor (F1) at b Probe
4 51
1 2
P PF
P P
5) Measure Velocity Calibration Factor (F2) at b
std2
1 2
PF
P P (
LDA1 22
VP P
2std LDA 2P V
![Page 12: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/12.jpg)
Prism Probe Calibration Results
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5 -50-40-30 -20-10 0 10 20 30 40 50
VLDA = 16.4 ft/s VLDA = 32.8 ft/s VLDA = 49.2 ft/s
VLDA = 65.6 ft/s VLDA = 82.0 ft/s VLDA = 98.4 ft/s
1F
a deg
![Page 13: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/13.jpg)
10 20 30 40 50
-1.0
0.0
1.0
-1.5
-0.5
0.5
1.5
-50-40-30 -20-10 0
1F
a deg
0.95
1.00
1.05
1.10
1.15
1.20
1.25
-50-40-30-20-10 0 10 20 30 40 50
2F
a deg
F2 for Prism
Probe has 5 %
Reynolds No.
Dependence
VLDA = 16.4 ft/s VLDA = 32.8 ft/s VLDA = 49.2 ft/s
VLDA = 65.6 ft/s VLDA = 82.0 ft/s VLDA = 98.4 ft/s
Prism Probe Calibration Results
5 %
![Page 14: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/14.jpg)
Spherical Probe Calibration Results
VLDA = 49.2 ft/sVLDA = 16.4 ft/s VLDA = 32.8 ft/s
VLDA = 98.4 ft/sVLDA = 65.6 ft/s VLDA = 82.0 ft/s
5
4
3
2
1
0
-1
-2
-3
-4
-5 -50 -40-30-20-10 0 10 20 30 40 50
1F
a deg
![Page 15: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/15.jpg)
10 20 30 40 50
VLDA = 16.4 ft/s VLDA = 32.8 ft/s VLDA = 49.2 ft/s
VLDA = 65.6 ft/s VLDA = 82.0 ft/s VLDA = 98.4 ft/s
Spherical Probe Calibration Results
1F
a deg
-5
-4
-3
-2
-1
0
1
2
3
4
5
-50 -40-30-20-10 0
2F
a deg
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
-50 -40-30-20-10 0 10 20 30 40 50
Spherical Probe
Reynolds No.
Dependence
at low pitch
![Page 16: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/16.jpg)
Applying 3-D Probe Calibration during RATA
Calibration
Parameters EPA Method 2F
NIST Implementation
of Method 2F
Pitch Calibration Factor (F1)
( a1 1F F ( a1 1F F ,Re,Tu
Velocity Calibration Factor (F2)
( a2 2F F ( a2 2F F ,Re,Tu
• EPA Method 2F; F1 and F2 are only functions of the pitch angle (a)
• 3-D probe calibration data showed the importance of accounting
for Reynolds number (Re) dependence
• NIST Implementation of Method 2F; F1 and F2 account for Reynolds
number (Re) and Turbulence (Tu) dependence
• Field Measured probe velocity
( probe 2 2 32 1 cos cos 0V F P P at P Pba
![Page 17: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/17.jpg)
Why is Turbulence Important?
• Wind tunnel probe calibrations are often performed in laminar
flow (i.e., turbulence intensity is nearly zero)
• Probes are used in stacks where flow is certainly turbulent
❖ Flow separation location and wake characteristics can vary significantly
between laminar and turbulent flow
❖ Pressure measurements located
In laminar-wake behind probe
will vary significantly from turbulent-wake
❖ Turbulent velocity fluctuations induce
an additional pressure at pressure ports.
(This turbulent induced pressure is not
present when the flow is laminar)
![Page 18: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/18.jpg)
How do we Generate Turbulence? Grid (12.5 cm spacing) Flag
• Turbulence intensity up to 11 % for grid and up to 25 % for flag
RMS2 2 2( ) 3u u v w
TuU U
• Turbulence intensity (Tu) is the rms of the velocity fluctuations
divided by mean velocity
• Magnitude controlled by downstream distance from grid or flag
![Page 19: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/19.jpg)
Does Turbulence Really Impact Accuracy?
Absolute Maximum Error
Root-Mean-Square Error
( probe 2 2 32 1 cos cos 0V F P P at P Pba
Prism Probe
F1 = F1(a) F1 = F1(a,Tu) F1 = F1(a,Re) F1 = F1(a,Re,Tu)
F2 = F2(a) F2 = F2(a,Tu) F2 = F2(a,Re) F2 = F2(a,Re,Tu)
EPA Method 2F
NIST Method 2F
Accuracy < 2 %
for -45° ≤ a ≤ 45°
% E
rro
r in
V pr
ob
e
5
4
3
2
1
0
![Page 20: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/20.jpg)
3 Non-Nulling Correlations
•
•
Measured
Parameters EPA
Method 007
NIST
Method 007
Normalized Pseudo
Dynamic Pressure
P1
PEPA
~ F( , ) ~
a b= Pt –
F( , , , ) ~
a b Re Tu PLDA
PNIST
~
=
Pitch Angle Function P4 – P5
PEPA
~ Fa ( , )a b= Fa ( , , , )a b Re Tu= P4 – P5
PNIST
~
Yaw Angle Function P2 – P3
PEPA
~ Fb ( , )a b= Fb ( , , , )a b Re Tu= P2 – P3
PNIST
~
~ PEPA = 0
Preliminary EPAMethod 007 will not work for 3-D
probes for which
NIST Method 007 o accounts for Reynolds (Re) number and Turbulence (Tu)
dependence P4
P3
P5
P2
P1
o works well for several probes over wide range of pitch and Prism Probe ~ yaw since ( PNIST > 0 )
![Page 21: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/21.jpg)
[deg]b
LDA
P
P
LDA
P
P
[deg]b
LDA
P
P
[deg]b
LDA
P
P
[deg]b
0.4
1.20a
1.2
0.4
21a
0.4
1.2
33a 45a
Non-Nulling: Example of Velocity Dependence Spherical Probe
16.4 ft/s 82.0 ft/s49.2 ft/s
32.8 ft/s 98.4 ft/s65.6 ft/s
-60 -40
0.6
0.8
1.0
-20 0 20 40 60 0.4
1.2
-60 -40
0.6
0.8
1.0
-20 0 20 40 60
1.2
0.4
-60 -40
0.6
0.8
1.0
-20 0 20 40 60 0.4
1.2
-60 -40
0.6
0.8
1.0
-20 0 20 40 60 0.4
1.2
![Page 22: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/22.jpg)
Custom Probe Shapes Designed at NIST
• NIST is researching various probe designs
• Probe performance is based on probe geometry and hole
placement
• Goal is to identify probes that
are highly immune to
Reynolds number effects and Turbulence over a wide range
of pitch and yaw
Cylindrical Probe Modified
Spherical Probe
Disk shape Truncated (flat)
Probe
![Page 23: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/23.jpg)
Scale-Model Smokestack Simulator (SMSS)
Air Intake
Unit
Air Exhaust
2 Fans Cone
➢ A Facility Designed to Assess the Flow Measurement Accuracy of CEMS and RATA
➢ Three Design Criteria 1) Facility must have CEMS and RATA equipment
commensurate to what is used in industry
2) Facility must create smokestack-like flow conditions
3) Facility must establish NIST traceable velocities (VNIST) to
compare CEMS and RATA
![Page 24: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/24.jpg)
CalibratedReferenceFlow Meter
SharpCorner
Scale-Model Smokestack Simulator (SMSS)
Air Intake
Module
Test Section
(Dtest = 4 ft)Exhaust
Test Probes (20 to 85 ft/s)
2 Fans
Calibrated Reference Flow Meter
Sharp Corner
1) 8 path ultrasonic flow meter measures flow to better than 0.5 %
2) Stack flow conditions (high swirl and skewed velocity profile) realized
by sharp corner section
3) RATAequipment installed in SMSS Test Section
RATA equipment – probes calibratedWind Tunnel installed in the automated
traverse system
![Page 25: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/25.jpg)
• Traverse arm rotated to RATA points on different
chords
❖ Completely Automated via LabVIEW software
Probe
RATA Performed using an Automated Pitot
Traverse Unit Installed in 4ft Test Section Rubber gasket
90 ° Orientation with metal straps
q
r b
0 ° Orientation
q
Pitot Traversing
Mechanism 0 ° Orientation
❖ Pitot probe can be positioned to any desired
location in the cross section Cross Section
• Probe moves radially to a selected RATA point
• Probe rotates to determine Yaw angle (b )
Probe b
![Page 26: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/26.jpg)
RATA Performed using an Automated Pitot
Traverse Unit Installed in 4ft Test Section Rubber gasket
q = 90° Cross Section ❖
90 ° Orientation
Probe
q
r b
0 ° Orientation
q
Pitot Traversing
Mechanism
with metal straps
Pitot probe can be positioned to any desired
location in the cross section
• Probe moves radially to a selected RATA point
• Probe rotates to determine Yaw angle (b )
• Traverse arm rotates to in q -direction to measur
RATA points on different chords
e
❖ Completely Automated via LabVIEW software
q q = 45°
![Page 27: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/27.jpg)
Test SectionReference
Meter
Location
Reference
Section
Dref
Dtest
x y
z
j
0°
90°
180°
270°
cross section
z
y
ExperimentalTraverseLocation
z
q
RATA Measurement Location
RATA performed
12 D from the Corner
Traverse Chord
z
yx
Prism Spherical S-probe Probe Probe
Cross section
![Page 28: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/28.jpg)
Test SectionReference
Meter
Location
Reference
Section
Dref
Dtest
x y
z
j
0°
90°
180°
270°
cross section
z
y
ExperimentalTraverseLocation
z
q
RATA: Velocity Profile, Yaw and Pitch Angles
Traverse
Chord
0 ° Orientation
Automated Probe Traverse
No
rma
lize
dA
xia
l V
elo
cit
y
y
z
Ya
w A
ng
le (
b)
Cross section
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0 -0.6
CFD
SP 2016
SP 2017
Prism
Sphere
-0.4 -0.2 0 0.2 0.4 0.6 y/D
Multiple Nulls found
close to the wall
40
30
20
10
0
-10
-20
-30 -0.6 -0.4 -0.2 0 0.2 0.4 0.6
y/D
![Page 29: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/29.jpg)
Test SectionReference
Meter
Location
Reference
Section
Dref
Dtest
x y
z
j
0°
90°
180°
270°
cross section
z
y
ExperimentalTraverseLocation
z
q
Inlet Cone
Pit
ch
An
gle
(a
)
.6 -0.4 -0.2 0 0.2 0.4 0.6
y/D
40
30
20
10
0
10
20
30 -0
RATA: Velocity Profile, Yaw and Pitch Angles
Traverse
Chord
0 ° Orientation
Automated Probe Traverse
y
z
No
rma
lize
dA
xia
l V
elo
cit
y
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0 -0.6
CFD
SP 2016
SP 2017
Prism
Sphere
-0.4 -0.2 0 0.2 0.4 0.6 y/D
Ya
w A
ng
le (
b)
-
-
-
Cross section
![Page 30: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/30.jpg)
S-Probe RATA along 2 Diametric Chords 45° Orientation 135° Orientation
# of Points VNIST, [ft/s] , [ft/s] VProbe % Difference
12 76.40 81.50 + 6.7 %
24 76.40 81.37 + 6.5 %
48 76.40 80.40 + 5.2 %
• In all cases the S-probe over predicts the actual flow
• Slight increase in accuracy with more traverse points
• Cp = 0.84; What is the accuracy if we use a calibrated S-probe?
![Page 31: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/31.jpg)
S-Probe RATA along 2 Diametric Chords 45° Orientation 135° Orientation
VNIST, [ft/s] # of Points VProbe, [ft/s] % Difference
12
24
76.40
76.40
81.50
81.37 + 6.5 %
+ 6.7 %
48 76.40 80.40 + 5.2 %
12 76.40 79.72 + 4.4 %
Cp =0.84
(Re,a)Cp
• In all cases the S-probe over predicts the actual flow
• Slight increase in accuracy with more traverse points
• Cp = 0.84; What is the accuracy if we use a calibrated S-probe?
• Calibration improves accuracy
![Page 32: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/32.jpg)
Prism Probe RATA along 2 Diametric Chords 45° Orientation 135° Orientation
VNIST, [ft/s] , [ft/s] # of Points VProbe % Difference
12 67.64 70.10 + 3.6 %
• The Prism probe over predicted the actual flow
• Better accuracy than calibrated S-probe (6.7 % uncalibrated)
![Page 33: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/33.jpg)
Spherical Probe RATA along 2 Diametric Chords
45° Orientation 135° Orientation
# of Points VNIST, [ft/s] VProbe, [ft/s] % Difference
12 67.63 68.52 + 1.3 %
• The Spherical probe over predicted the actual flow
• Better accuracy than S-probe (6.7%) and the Prism probe (3.4 %)
![Page 34: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/34.jpg)
non-
over
Non-Nulling Method Works Well!
• NIST has developed a robust, high accuracy
non-nulling method
o Improvement over Method 007 (i.e., more accurate fit
over wide range of pitch and yaw)
o Accounts for turbulence (Tu) and Reynolds number (Re)
dependence
• Recap Spherical Probe RATA Results
o Measured VRATA and VNIST
o Accuracy evaluated by ( RATA NIST% Di V Vff 100 1
# Points % Diff
12 + 1.3 %
Method Tu
(During Cal.) Tu
(During Use)
2F 0 % 10 %
Non-Null 10 % 10 % - 0.8 % 12
• Results are preliminary pending field test
![Page 35: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/35.jpg)
Summary 1) Wind Tunnel Probe Calibrations
❖ S-probe has a large pitch dependence (10 % effect) that cannot be
accounted for via calibration.
❖ 3-D Probes highly accuracy if Reynolds dependence and Turbulence
characterized
❖ Robust non-nulling techniques have been developed and work well!
❖ New probe designs less sensitive to turbulence and Reynolds number
are being developed
2) SMSS Facility Results
❖ RATA Testing
o Spherical probe exhibited best accuracy (± 1 %)
o prism probe (~ +3 %)
o S-probe (~ + 6 %)
❖ SMSS Facility has large yaw angle ~35° near the wall
❖ Accuracy of Non-nulling method is the same as yaw-nulling method
(within 1 %)
![Page 36: NIST’sWind Tunnel and Stack Simulator](https://reader033.vdocument.in/reader033/viewer/2022042013/6255b24840dcab0e206b1804/html5/thumbnails/36.jpg)
Questions?