1. Agricultural and Biological Engineering
Open Path Methods
Albert J. Heber, Professor
Building Environment
Research & Education
AgAirQuality.com
Agricultural and Biological Engineering
Purdue University
Slides: Title: Heber only under title

2. Biocurtain at Laying House
Biocurtain over 3 fans
Lab

3. Inside Biocurtain at Layer House

4. NRC Report on Air Emissions
Global/Nat.
Local
Concern
NH3
Major
Minor
N-Dep/PM2.5
N2O
Significant
Insignificant
Climate
NOx
Haze/Health
CH4
VOCs
Quality
H2S
PM10
Haze
PM2.5
Health/Haze
Odor

5. Comparing Open Path Sensors
Type of Sensor
FTIR UV
OPL
Detector cooling
Cryocooler -
Path Length, m
400
2000
Mode
Monostatic
Bistatic
Compounds
NH3, VOC*, CH4
NH3, H2S, Nox
H2S or NH3
Scan frequency, Hz 1
4000
Detection Limits
Hydrogen sulfide
Deuterium
ppm-m
10-30
0.4-5
6-25
ppb
75-600
2.8-33
3-120
Linear upper range
Ammonia
Xenon
1.5 2
2-50
3-20
1-40
903
Models
BLS, TOM
BLS

6. Repump cooler, replace retros
Type of Sensor
FTIR UV
OPL
Scanning
Yes ?
Reflectors -
200 m
30 cube
Small retro
400 m
60 cube
1000 m
90 cube
Real-time quantification
yes
yes, w/ BLS
Capital cost
140K
$20K-$45K
$30K
Short term costs
none
Annual costs
Repump cooler, replace retros
New source
None
Recalibration needs
Annually
$7K laser/7 yr.

7. FTIR with 48-m Closed Cell
Advantages
Measures greenhouse gases
Measures ammonia: MDL=<6 ppb, NO, NO2
Measures dozens of other gases, SO2
Real-time measurement
Quick response: limited by cell volume
Disadvantages
Expensive: $75,000
Heavy, non-portable
Slides – FTIR Nexus 670: Advantages and Disadvantages

8. Scanning FTIR - Tomography
Layer house
Horizontal scanning
Vertical scanning

9. Source: Bruce Harris, U.S. EPA, 2004

10. Source: Bruce Harris, U.S. EPA, 2004

11. Field Measurement of Air Pollutants Near Swine Confined Animal Feeding Operations using UV DOAS and FTIR C. D. Secrest (paper presented in 2000)
Ambient ammonia concentrations 0.8 km from a large swine facility with lagoons over a two week period were 0 to 900 ppb.
An Iowa Study Group recommended that ambient exposure to ammonia should not exceed 150 ppb.
The UV DOAS and FTIR were in good agreement.
Open-path monitors combined with wind monitors are powerful tools for comparing daytime and nighttime pollutant concentrations, and for determining the effect of wind speed on concentration.

12. Area sources -> diffuse plumes Open-path -> entire plume length
As far as the detection limits:  One FTIR manufaturer suggests a noise limited detection limit of 10 ppm-m for H2S. This yields a 25 ppb limit for an optical path length is 400m (200m physical). In reality, at least 3 times that should be used as the detection limit. It may be as much as 200 ppb for real world situations. The UVDOAS at 5 ppm=m DL calculates to 25 ppb at 200 m. I don't know if the UV is a noise limited number or not.  Both of these calculations do not take into account interferences. The water may play havoc with the FTIR , while other atmospheric absorptions may bother the UVDOAS. At 200 nm, the O2 aborption limits transmission to 0.67% of the signal.       The 'jump the fence' problem would be the same for the FTIR or the single beam as the lower horizontal beam in the RPM method coincides with the single UV beam. With the possible option to lengthen the paths to provide more vertical or horizontal definition (There is a limit of probably m.), the unit could be backed away from the edge of the source to better capture the boundary of the plume.       Of all the assumptions we've been making, the one that inpacts greatly upon our calulations is that of uniform emission from the surfaces being measured. Our data for lagoons and landfills show that this is not the case most of the time. Regardless of what one takes from Lori Todd's work on the Farm 10 lagoon, her traditional intersecting CT scan showed large variations across the surface. Everbody gets around this by using a path average. Perhaps the vertical extent of the plume is also not uniform, We have seen this many times. Does this impact our calculations? We try to limit the vertical question by using elevated beams and the horizontal by breaking the ground beam into segments. The limit on height is usually the available lifting mechanism or other structure to which the retros are attached. The closer we can get to the source the better chance we have of capturing the plume. Our whole concept is to measure the plume at multiple points to establish a concentration profile as is done in EPA stack methods and integrate it with the flow field to develop a measured flux.   (Once a stack sampler, always a stack sampler!) The closest thing to a model in our work is the interpolation of the concentration profile in the plume. We continue to try to improve this methology and have additional work underway. We're hoping that the 3D anemometer will improve the flow field information over the 2 & 10 meter boundary layer assumption.       An interesting possibility which one manufacturer has presented is to combine UV & FTIR in one unit. It has been built yet and shouldn't be considered, but does show how fast the instrument manufacturers can rise to new opportunities. Bruce
Area sources -> diffuse plumes
Open-path -> entire plume length
An array paths maps the plume
Source: Bruce Harris, U.S. EPA, 2004

13. OP-FTIR Measurement Paths for Path-Integrated Optical Remote Sensing (Tomography)
wind
Al,
The computed tomography (CT) method of estimating the plume structure, as reported by Hasmonay, Harris, and Yost and others in EST 2001, 35: , while not strictly a gaussian plume model, relies on the ensemble-averaged homogeneous turbulence assumption within an Eulerian field that is inherent in the traditional Gaussian plume model. The CT approach to defining the plume cross-sectional domain area necessarily requires assumptions concerning the shape of the plume. Their smoothed basis function minimization (SBFM) reconstruction of the plume relies on a guassian distribution of concentration of a single plume based on the measured sigmas to extrapolate and interpolate the entire plume cross-section from the relatively small fraction of measured slices of the plume made along the FTIR scan lines.
Also inherent in the method is the assumption that they have determined the plume location in Z along the scans (hence several intermediate distances for scans at the surface). It is possible that the method would not work well if the plume centerline did not fall within the measured scan area- as may occur if we measure the plume away from the pond by say 10m and be on the other side of the berm or may occur due to the bouyancy of air over thae pond during periods when the pond is warmer than the surroundings.
Consequently this approach should have the same limitations of near-field utility that the tradiational Gaussian models have.
Richard H Grant
Prof. Applied Meteorology
Dept of Agronomy, Purdue University
915 W. State St.
W. Lafayette, IN
Source: Bruce Harris, U.S. EPA, 2004

14. Controlled release simulation of an area source under unstable air conditions – worst case40 80
120
160
200
240 2 6 10 14
Crosswind Distance [meters]
Height [meters]
Oxford 10/15/99: average flux g/s
0.2
0.4
0.6
0.7
concentrations are in mg/m 3 40 80
120
160
200
240 2 6 10 14
Crosswind Distance [meters]
Height [meters]
Oxford 10/15/99: Run #1 flux g/s
0.1
0.3
0.4
0.5
concentrations are in mg/m 3
Reconstructed plumes
Actual release rate = 1.7 g/s
Calculated flux = 1.2 g/s
Measured σθ = 50.7°
Pasquill-Gifford Stability A - Unstable 40 80
120
160
200
240 2 6 10 14
Crosswind Distance [meters]
Height [meters]
Oxford 10/15/99: Run #2 flux g/s
0.2
0.3
0.5
0.6
concentrations are in mg/m 3
Height [meters] 40 80
120
160
200
240 2 6 10 14
Crosswind Distance [meters]
Height [meters]
Oxford 10/15/99: Run #3 flux g/s
0.2
0.3
0.5
0.7
concentrations are in mg/m 3
Source: Bruce Harris, U.S. EPA, 2004

15. Controlled release simulation of an area source under stable air conditions – best case40 80
120
160
200
240 2 6 10 14
Crosswind Distance [meters]
Height [meters]
Oxford 10/19/99: average flux g/s
0.9
1.8
2.6
3.5
concentrations are in mg/m 3
Oxford 10/19/99: Run #1 flux g/s
1.9
2.8
3.8
Oxford 10/19/99: Run #2 flux - 1.6g/s
1.1
2.1
3.2
4.2
Oxford 10/19/99: Run #3 flux g/s 7
10.5
Oxford 10/19/99: Run #4 flux g/s 1
2.9
3.9
Reconstructed plumes
Actual release rate = 1.7 g/s
Calculated flux = 1.5 g/s
Measured σθ = 12.7°
Pasquill-Gifford Stability C-D - Neutral
Source: Bruce Harris, U.S. EPA, 2004

16. FTIR References
Harris, D. B., and E.L. Thompson, Jr Evaluation of ammonia emission from swine operations in North Carolina. Proc. Emission Inventory-Living in a Global Environment, VIP-88, pp AWMA, Pittsburgh, PA.
Harris, D. B., E.L. Thompson, Jr., D.A. Kirchgessner, J.W. Childers, M. Clayton, D.F. Natschke, W.J. Phillips Multi-pollutant concentration mapping around a concentrated swine production facility using open-path FTIR spectrometry. Workshop on Atmospheric Nitrogen Compounds II: Emissions, Transport, Transformation, Deposition and Assessment, NCSU, Raleigh, NC, pp
Childers, J. W., E.L. Thompson, Jr., D.B. Harris, D.A. Kirchgessner, M. Clayton, D.A. Natschke, W.J. Phillips Multi-pollutant measurements around a concentrated swine production facility using open-path spectrometry. Atm. Env. 35:
Childers, J. W., Thompson, E. L., Jr., Harris, D. B., Kirchgessner, D. A., Clayton, M., Natschke, D. A., Phillips, W. J. (2001) Application of standardized quality control procedures to open-path fourier transform infrared data collected at a concentrated swine production facility. Env. Science & Tech. 35:
Source: Bruce Harris, U.S. EPA, 2004

17. FTIR References
Childers, J. W., E.L. Thompson, Jr., D.B. Harris, D.A. Kirchgessner, M. Clayton, D.A. Natschke, W.J. Phillips Comparison of an innovative algorithm to classical least squares for analyzing open-path fourier transform infrared spectra collected at a concentrated swine production facility. Appl.Spect. 56:
Hashmonay, R. A., D.A. Natschke, K. Wagoner, D.B. Harris, E.L. Thompson, Jr., M.G. Yost Field evaluation of a method for estimating gaseous fluxes from area sources using open-path fourier transform infrared. Env. Sci. Tech. 35:
Harris, D. B., E.L. Thompson, Jr., Vogel, C. A., Hashmonay, R. A., Natschke, D. A., Wagoner, K. Yost, M.G. Innovative approach for measuring ammonia and methane fluxes from a hog farm using open-path fourier transform infrared spectroscopy. 94th Annual Conf. of the AWMA, VIP-102-CD, AWMA, Pittsburgh, PA 2001.
Hashmonay, R.A. and D.B. Harris Particulate matter measurements using open-path Fourier transform infrared spectroscopy. 94th Annual Conference of the Air & Waste Management Association, VIP-102-CD, AWMA, Pittsburgh, PA.
Harris, D.B., R.C. Shores, L.G. Jones. Ammonia Emission Factors from Swine Finishing Operations. Int. Emissions Inventory Conference, “One Atmosphere, One Inventory, Many Challenges.”
Source: Bruce Harris, U.S. EPA, 2004

18. Neutral Stability ponds sheds ponds sheds Source: Lowry Harper
Notice that a complete flux measurement (flux sampling of plume to large height) overestimates the flux by ~ 10% (this is because of the neglect of turbulent flux)
15 m
5 m
25 m
C plume in neutral conditions along measurement plane
Flux measurement plane
Neutral Stability
Wind
ponds
sheds
Hi Folks,
Here are my ideas, in collaboration with my research colleagues, on the use of a mass-balance technique for real farm situations:Attached you'll find a PowerPoint file with three graphs which detail how a mass-balance measurement might look like at our actual study farm in Utah. I felt this might be a good example of the scale of tracer source which might be of interest. More on this below.
1. A mass-balance approach to measuring fluxes is probably the best technique, in theory. It is "model-free" with few assumptions. We have used the technique many times with great success (see previous development uses in prior correspondence), and if the field situation allows it, it would be our preferred method. The only theoretical downside is that if the flux is represented by the product of a mean concentration <C> and a mean windspeed <U> (where the angle brackets represent a time average):
FLUX_IHF = <U><C>
then there is an error due to a neglect of the turbulent flux. This is because the flux is really
FLUX = <UC> = <U><C> + <U'C'>
The error by neglecting <U'C'> is probably 5-15% (overestimation) depending on the measurement location and the atmospheric stability (there is considerable discussion on this in the literature).
2. In general, the mass-balance approach works well for small sources, but not large ones. We have used the technique quite successfully for measuring enteric emissions. This is because of the scale of the plume which must be sampled. Have a look at PowerPoint graphs developed by Tom Flesch. We have considered the Utah nursery farm we studied, and assumed we wanted to make a mass balance measurement. In the case of this actual farm, you'd have to position your measurement plane at least 50 m downwind of the farm complex to get away from the berm and get to level ground.
In the attachment, you can see the extent of the plume in neutral, unstable, and stable atmospheric conditions. The graph shows a long swine house and two adjacent lagoons. The red line shows the placement of a measurement (tomographic?) plane up to a height of 50 feet in the three slides. The picture of a simulation of the plume in vertical cross section shows what the plume would like emitting from the lagoons and barn at the location of the red plane. The X-Y graph shows what the emissions would be (ratio of the measured or calculated flux to the actual flux) heights up to 50 meters.
In neutral conditions (windy, cloudy weather) the tracer will be organized in two distinct plumes (barn and lagoons). The plume will extend to approximately z = 20 m. Observations up to a height of ~ 15 m would be required to get the flux to within 10%. Notice that a complete measurement of the plume yields a 10% overestimate of the flux (this is due to a neglect of the turbulent flux discussed above).
In unstable conditions (L = -5m, daytime, sunny, light winds) the farm plume is more smeared, with only one distinct plume. The vigorous turbulence has dispersed the plume much more than in neutral conditions. Here the plume extends to a height of 50+ m. Observations above 50 m are required to measure the flux to within 10%. Notice that although the concentration is low at z=50m, the windspeed is high, so the flux is substantial. It would be extremely difficult to measure high enough in this climatic condition–which occurs a large percentage of the time.
In stable conditions the mixing is much reduced, and the plume only extends to about z = 5m.
Given the spatial extent of the plume in the neutral and unstable cases, it would seem that a mass balance approach would be too costly and troublesome for a farm of this scale (i.e. ~ 100 m in horizontal extent). We have found in numerous small-scale studies with reasonably uniform fetch (no buildings) we had to have a length to measurement height ratio of 100/15 (i.e. for every 10 meters of length, we need to measure up to 15 meters–and as stated above, the height may need to be higher).
3. Notice that for the Utah farm complex the plume is "multifaceted" in the vertical plane (i.e. has more than one C maximum). Our intuition suggests that it would take many laser measurements of CL to deduce this pattern (clearly the more detail you need to deduce, the more concentration observations will be required). I would think that in this case they would have to not only have concentration measurements across the full span of the plane, but would need partial spans as well.
And finally, there is no doubt in our opinions that a mass-balance measurements for real-sized farms will be a costly and difficult task. It is worth noting that the bLS method was originally conceived because of the effort required to make a mass balance measurement. Lowry Harper
Source: Lowry Harper
USDA-ARS, 2004
ponds
sheds

19. Unstable (daytime) Smeared plume
C plume in unstable conditions along measurement plane
Flux measurement plane
Notice that even if you go to z=50 m you don’t capture all the flux
Unstable (daytime)
Wind
In unstable conditions (L = -5m, daytime, sunny, light winds) the farm plume is more smeared, with only one distinct plume. The vigorous turbulence has dispersed the plume much more than in neutral conditions. Here the plume extends to a height of 50+ m. Observations above 50 m are required to measure the flux to within 10%. Notice that although the concentration is low at z=50m, the windspeed is high, so the flux is substantial. It would be extremely difficult to measure high enough in this climatic condition–which occurs a large percentage of the time.
Source: Lowry Harper
USDA-ARS, 2004
Smeared
plume

20. 15 m
5 m
25 m
C plume in stable conditions along measurement plane
Flux measurement plane
Stable (nighttime)
Wind
In stable conditions the mixing is much reduced, and the plume only extends to about z = 5m.
Given the spatial extent of the plume in the neutral and unstable cases, it would seem that a mass balance approach would be too costly and troublesome for a farm of this scale (i.e. ~ 100 m in horizontal extent). We have found in numerous small-scale studies with reasonably uniform fetch (no buildings) we had to have a length to measurement height ratio of 100/15 (i.e. for every 10 meters of length, we need to measure up to 15 meters–and as stated above, the height may need to be higher).
Source: Lowry Harper
USDA-ARS, 2004

21. Backward Lagrangian Stochastic (BLS) Dispersion Models
Backward Lagrangian Stochastic Modeling
Introduced by Flesch, T.K., and J.D. Wilson Backward-time Lagrangian stochastic dispersion models and their application to estimate gaseous emissions. J. Applied Meteorology 34:
Utilizes point or line measurement
Ultrasonic or cup anemometers
Flexible and easy to use.
Surface layer model. Locate < 1 km.
Commercial software available
Dear Group, In an effort to bring some perspective to the Backward Langrangian Stochastic (BLS) modeling techniques being discussed for the Consent Decree monitoring protocol, my colleague, Prof. Pal Arya and I reviewed Flesch et al. (2003) (“Deducing Ground Air Emissions from Observed Trace Gas Concentrations: A Field Trial.” A copy attached). In this paper which has been cited in our discussions, the authors discuss a field experiment where BLS was used to model emissions from a small area (6m X 6m) under ideal conditions. This review (see attachment) indicates that, under the conditions likely to be encountered around livestock waste lagoons, variance in emission estimates and costs to conduct the measurement would both be high. Some of the major limitations of inverse dispersion modeling, including the BLS model of dispersion in the atmospheric surface layer from animal waste lagoons at conventional swine and other animal farms in the U.S., are also pointed out in the review. I would be glad to discuss these findings. Thanks. Viney.
Al,
There are several issues here:
I have looked over the Aneja/Arya evaluation of the backward Lagrangian stochastic method vs. the Gaussian method. The uncertainty in emission estimates are similar for a given distance from the source as indicated in the critique, but bLS should perform better in the near field since the diffusion processes (due to small scale turbulence and well-modeled by a Gaussian models) are overwhelmed by the dispersion processes (due to larger scale turbulence, and best modeled by Lagrangian models) in short time periods- resulting in more advection than diffusion of the plume.  This is why diffusion/dispersion in stable atmospheres are harder to model than unstable atmospheres.  All model fail under these conditions due to the minimal diffusion and dominant dispersion.
Measuring close to the source minimizes deposition issues presented in the critique and for the above rationale almost dictates the use of bLS.  However too close will not work due to the short pathlengths of air parcels being in the extreme low end of the stochastic probability functions.
We reduce the modeling error for either the Gaussian or bLS method by constraining the final emission estimate through the use of multiple paths through the ‘plume’ (and consequently multiple, replicate, concentration samples at different location in the plume.  However I have not seen any studies indicating the degree of improvement on the emission estimate.  
Richard H Grant
Hi Folks, In the conference call I was asked to provide references to the bLS technology. There are several sources: 1. Please look in the section addressing the bLS in the book chapter that I sent today. There is a brief discussion of the technique in the chapter describing it, its advantages and disadvantages, and some potential uses. Literature citations are in the section. 2. An in-depth discussion of the bLS technique is given in a book chapter, also in the same Monograph, by Tom Flesch (Ph , Further references are surely given there but I haven't seen a draft copy of his chapter in some time. Please call or him if you would like this information or perhaps a pdf file of his chapter (and maybe Jerry won't get too upset). 3. Further work has been done (superseding the previous papers) on the technique since these publications. A field study using NH3 and CH4 as tracer gases compared released gases to bLS evaluations and the paper is in press. The reference will be Flesch, T.K., Wilson, J.D., Harper, L.A., Crenna, B.P., and R.R. Sharpe. Deducing Ground-air emissions from observed trace-gas concentrations: A field trial. Am. Meteorol. Soc. (In press) A preprint copy of the paper is attached as a pdf file. 4. Further field studies have been completed and the manuscript on these studies is in review. This study gives the bLS use in evaluating a whole-farm emission rate for NH3 and CH4. We have no tracer emissions for absolute comparison but the results are similar to emissions we have found in the measurement of combined individual measurements of these gases (not at the same time). This paper reference will be Flesch, T.K, J.D. Wilson, and L.A. Harper. Estimating Farm Emissions of Ammonia and Methane with an Inverse-Dispersion Technique. Am. Met. Soc. (in review). I cannot send a pdf file of it as the senior author is not ready to release it. 5. The bLS technique and the mathematics have been put into a Windows-type software package that has recently become available commercially. Our original studies were done on our own (Tom and John) programs but now we are using the new available software for all our studies--it is much easier to use. You may find information about the software at the following website: . If you have any questions, please me. Best regards--Lowry Lowry A. Harper Agricultural Research Service United States Department of Agriculture 1420 Experiment Station Road Watkinsville, GA Ph: x225 fax:

22. UV-DOAS Ultraviolet Differential Optical Absorption Spectroscopy
ppb path length
Fast scanning, compact, tunable
EPA Equivalent Method for SO2, O3 and NO2.
Also measures ammonia, benzene, toluene, xylenes, styrene, Hg, HF, HNO2, HCHO
Continuous operation
MDL for ammonia = 2.8 to 5.8 ppb
Source: Myers, J., T. Kelly, C. Lawrie, and K. Riggs ETV Technology Evaluation Report. Opsis, Inc. AR-500 Ultraviolet Open-Path Monitor. ETV Advanced Monitoring Systems Center, Battelle.

23. EPA Lab for Ambient Measurements
TEOM
UV-DOAS
UV-DOAS
TEOM
UV-DOAS
1-min averaging and
recording intervals
MET
tower

24. Collocated UV’s

25. Micromet Setup at Lagoons
FTIR & Tomography
UV & BLS
FTIR & BLS
Source: Bruce Harris, U.S. EPA, 2004

26. Equipment Required per Team
Two FTIR scanning systems with 20 retros
Two UV systems
Four computers for optical remote sensors
One computer for data QAQC and analysis
Two 3D ultrasonic anemometers (2 and 12 m)
Complete weather station
Two, 12-m towers for FTIR/UV systems
One, 2 m tower for ultrasonic anemometer
Software for computed tomography method
Software for BLS method
Van and trailer

27. Check out AgAirQuality.com
Thank you!
Slides – Thank you: Haven Acres
Check out AgAirQuality.com