1. Integrating Modeling and Physical Testing for Assessing Filtered Exhaust Stack Sampling Probe Location
Xiao-Ying Yu, Kurtis P. Recknagle,
John A. Glissmeyer, & J. Matthew Barnett
Health Physics Society Annual Meeting
Air Monitoring Session MPM-E
Pacific Northwest National Laboratory
Richland, WA
July 14, 2014
PNNL-SA

2. Introduction 3410 Building at PNNL exhaust system was re-configured
Exhaust capability increased
Additional fan added to exhaust system
New ANSI/HPS N testing was required to establish a well-mixed location for sampling equipment
Computational Fluid Dynamics (CFD) was used as a predictive model
Actual full-scale tests were conducted
Results of CFD modeling and Full-scale testing are reported
July 14, 2014
PNNL-SA

3. Re-qualification of Modified Stack
3410 Building stack sampling location had previously been qualified by similarity to a tested scale model
Stack configuration was changed with the addition of a third fan
Previous qualification was invalidated
Redesigned stack needed to be qualified anew
July 14, 2014
PNNL-SA

4. ANSI/HPS N13.1 Mixing Criteria
Uniform Air Velocity: The air velocity uniformity expressed as the Coefficient of Variance (COV) must be ≤ 20% across the center two‑thirds of the area of the stack
Angular Flow: Sampling nozzles are usually aligned with the axis of the stack. The average air velocity angle must not deviate from the axis of the stack or duct by more than 20°
Uniform Concentration of Tracer Gases: Tracer gas is injected downstream of the fan and also at the junction of the ducts. The COV of the measured tracer gas concentration must be ≤ 20% across the center two-thirds of the sampling plane; and at no point in the sampling plane does the concentration vary from the mean by > 30%
Uniform Concentration of Tracer Particles: Tracer particles of 10-μm aerodynamic diameter (AD) are used. The COV of the measured particle concentration must be ≤ 20% across the center two‑thirds of the sampling plane
July 14, 2014
PNNL-SA

5. Important Technical Details in Physical Testing
Pattern and dimensions for wall injection points during gas tracer testing
Cross-section of the duct at the testing ports with measurement points
July 14, 2014
PNNL-SA

6. Modeling Approach
Approach included first using CFD modeling to determine the likely success of physical full-scale qualification testing
CFD model was predictions were compared to the previous test’s air velocity profile for model validation
CFD modeling was used to optimize probability of passing full-scale testing
July 14, 2014
PNNL-SA

7. 3410 Building Stack Operating Parameters
New
Old
Duct diameter at sampling probe (in.) 40
No. of duct diameters from upstream disturbance to sampling probe
abt. 7 10
No. of duct diameters from sampling probe to downstream disturbance 3
Discharge diameter (in.) 32
Single fan capacity (actual ft3/m, acfm)
21,200
No. of operating fans 2 1
Total available fans
Maximum flow rate (acfm)
42,400
19,900
Operating flow rate (acfm)
31,840
17,000
Reynolds number
1.42E+05
6.69E+04
Air temperature at exit (°F)
40–75
July 14, 2014
PNNL-SA

8. Modeling Results
The CFD modeling indicated the proposed stack configuration would fail to meet the mixing criteria
Particle distributions at the sampling section for a total modeled flow rate of 31,840 cfm without air blender
a) fans 1 and b) fans 1 and c) fans 2 and 3
July 14, 2014
PNNL-SA

9. Modeling Results a) fans 1 and 2 b) fans 1 and 3 c) fans 2 and 3
Tracer gas concentration contours at the sampling section for a total modeled flow rate of 31,840 cfm without air blender
a) fans 1 and b) fans 1 and c) fans 2 and 3
July 14, 2014
PNNL-SA

10. Modeled Results at Sampling Station
Mid-duct injection
Fans operating
Flowrate, cfm
Particle COV, %
Tracer gas COV, %
1 & 2
31,840 36
6.2
1 & 3 50 24
2 & 3 58 13
July 14, 2014
PNNL-SA

11. Modeled Results at Sampling Station
Fan injection
Fans
Inject location
Total Flow, cfm
Velocity COV, %
Particle COV, %
Tracer gas COV, %
1 & 2
Fan 1
31,840
4.7 29
Fans 1, 2 16
4.1 1
15,920
6.3 18
5.0
July 14, 2014
PNNL-SA

12. Modeling Results
An in-stack air blender was modeled as a potential solution 1 2 3
July 14, 2014
PNNL-SA

13. Static Air Blender S40C3S Photo of the air blender
Computational mesh used in CFD
July 14, 2014
PNNL-SA

14. Modeling Results
Tracer particle distributions at several stations downstream of the air blender showing good mixing
July 14, 2014
PNNL-SA

15. Modeling Results
CFD mixing results with air blender show when the mixing criteria are met
July 14, 2014
PNNL-SA

16. Summary Full-Scale and CFD Modeled Results
Parameter
Average Full-Scale Result
Average CFD Modeled Result
Uniform Air Velocity COV
2.9%
2.5%
Angular Flow
4.0
1.5
Uniform Concentration of Tracer Gases COV
Avg. 2.8%; max deviation < 6.7%
1.3%
Uniform Concentration of Tracer Particles COV
<9.9%
16%
July 14, 2014
PNNL-SA

17. Conclusions
CFD modeling was a useful tool to assess the likelihood of successful physical testing of a reconfigured stack exhaust system
CFD modeling identified a poorly-mixed sampling location with the proposed new configuration
Additional CFD modeling indicated an air blender would alleviate the poor mixing
The final configuration included an air blender, and physical tests successfully met the ANSI/HPS N13.1 mixing criteria for the sampling location
July 14, 2014
PNNL-SA

18. Thank you!
Contacts: John Glissmeyer: Matthew Barnett:
July 14, 2014
PNNL-SA

19. Additional Slides
July 14, 2014
PNNL-SA

20. Summary of Tracer Gas Uniformity Tests
Fans
Run No.
Injection Port††
Injection Position
Max Dev.
COV, %
Fan 1 & 3
GT-12
Junction Center
Center and 4 Perimeter Points
-1.9%
1.2
GT-13
Junction Top
1.6%
0.9
GT-14
Junction Bottom
1.7%
GT-15
Junction Near
2.6%
1.5
GT-16
Junction Far
1.4%
0.7
GT-17
Fan 3 Center
1.5%
0.8
GT-18
Fan 3 Top
GT-19
Fan 3 Bottom
0.6
GT-20
Fan 3 Far
-2.0%
0.5
GT-21
Fan 3 Near
-2.9%
1.9
GT-22
Fan 1 Center
Center
-3.0%
1.1
GT-44
near
6.7%
2.8
†† All physical tests were performed after the installation of the air blender.
July 14, 2014
PNNL-SA

21. Summary of Particle Uniformity Tests
Fans
Injection Port††
Normalized COV, %
Fans 1 & 2
Junction
8.28
Fan 1
8.93
Fan 2
7.77
Fans 2 & 3
7.01
8.53
Fan 3
9.56
Worst Case (2 & 3)
6.52
9.87
9.16
8.46
Worst Case (1 & 3)
7.74
†† All physical tests were performed after the installation of the air blender.
July 14, 2014
PNNL-SA