- Email: [email protected]
Direct measurements of the ozone formation potential from dairy cattle emissions using a transportable smog chamber
ARTICLE IN PRESS
Atmospheric Environment 42 (2008) 5267–5277 www.elsevier.com/locate/atmosenv
Direct measurements of the ozone formation potential from dairy cattle emissions using a transportable smog chamber Cody J. Howarda, Wenli Yangb, Peter G. Greena, Frank Mitloehnerc, Irina L. Malkinac, Robert G. Flocchinib, Michael J. Kleemana, a
Department of Civil and Environmental Engineering, University of California at Davis, 1 Shields Avenue, Davis, CA 95616, USA b Crocker Nuclear Laboratory, University of California at Davis, 1 Shields Avenue, Davis, CA 95616, USA c Department of Animal Science, University of California at Davis, 1 Shields Avenue, Davis, CA 95616, USA Received 15 August 2007; received in revised form 8 February 2008; accepted 28 February 2008
Abstract Tropospheric ozone continues to be an air pollution problem in the United States, particularly in California, Texas, and across the eastern seaboard. The obvious sources of ozone precursors have been largely controlled over the past several decades, leading to the critical examination of secondary sources. In particular, California has new air quality rules addressing agricultural sources of ozone precursors, including dairy farms. Some recent estimates predict that dairy cattle are second only to on-road vehicles as a leading source of ozone precursor emissions in California’s San Joaquin Valley. The objective of this work was to directly measure the ozone formation potential from dairy housing. A transportable ‘‘smog’’ chamber was constructed and validated using organic gases known to be present in dairy emissions. The ozone formation potential of emissions from eight non-lactating dairy cows and their fresh waste was then directly evaluated in the field at a completely enclosed cow corral on the campus of the University of California, Davis. The results demonstrate that the majority of the ozone formation is explained by ethanol (EtOH) in the emissions from the dairy cows, not by acetone as previously thought. Ozone formation potential is generally small, with o20 ppb of ozone produced under typical conditions when EtOH concentrations were 200 ppb and NOx concentrations were 50 ppb. The results match our current understanding of atmospheric ozone formation potential, ruling out the possibility of unknown organic compounds in dairy emissions with significant ozone formation potential. Simulations carried out with a modified form of the Caltech Atmospheric Chemistry Mechanism verify that actual ozone formation from dairy emissions is much lower than what would be predicted using the current regulatory profiles. Based on these results, the ozone formation potential of emissions from dairy cattle in California seems to be lower than previously estimated. r 2008 Elsevier Ltd. All rights reserved. Keywords: Ozone formation potential; Portable smog chamber; Ethanol; Acetone
1. Introduction
Corresponding author. Tel.: +1 530 752 8386;
fax: +1 530 752 7872. E-mail address: [email protected] (M.J. Kleeman). 1352-2310/$ - see front matter r 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2008.02.064
Ground level ozone continues to be a problem throughout the United States (US Environmental Protection Agency, 2007) and particularly in California’s San Joaquin Valley (SJV) and South Coast
ARTICLE IN PRESS 5268
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Air Basin (SoCAB). Although mobile sources still account for the large majority of reactive organic gases (ROG) (photochemically reactive and not exempt as an ozone precursor), California has made great strides in decreasing those emissions and has begun looking for other sources, which could lead to ozone formation. One such source that has recently come under scrutiny is livestock waste, specifically from dairy cattle. Dairy is the dominant livestock industry in California, where 1.78 million dairy cows produced 38.8 billion pounds of milk in 2006, generating 21% of the national supply (Department of Food and Agricultural-California, 2007). According to the recently approved ozone plan for the SJV, livestock waste from dairy cattle account for 17.6 ton of volatile organic compounds (VOC, organic compounds which readily evaporate to the atmosphere) emissions per day and 16% of the ROG in the SJV (San Joaquin Valley Air Pollution Control District, 2007). Only mobile sources contributed more to the total ROG emitted in the region. Based on these data, the San Joaquin Valley Air Pollution Control District (SJVAPCD) approved a plan which aims to decrease VOC emission by 15.8 ton day1 from all concentrated animal feeding operations (CAFOs), with dairy farms accounting for 80% of all CAFOs (San Joaquin Valley Air Pollution Control District, 2006). The SJVAPCD based their estimates for the ozone forming potential (OFP) of dairy cow emissions on research performed on total organic gas (TOG) emissions from cows in the 1930s (Ritzman and Benedict, 1938) and an erroneously determined ROG/TOG factor (US Environmental Protection Agency, 1980; Mitloehner, 2005). Recent findings suggest that ROG emissions from dairy farms may be much lower than previously speculated and the potential for ozone formation from these sources may therefore be less than anticipated. According to Shaw et al., (2007), ROG fluxes from dairy cattle are approximately 6–10 times lower than the value suggested by present emission estimates. They calculated an ozone formation potential (OFP) of dairy cattle emissions that is 10% lower than that from combustion and biogenic sources (Shaw et al., 2007). They base this result on three separate OFP model calculations of ethanol and methanol (Russell et al., 1995; Derwent et al., 1998; Hakami et al., 2004), which are reported to account for 80–90% of the ROG in the dairy emissions and make up 1.8% of the TOG emissions
(Shaw et al., 2007). Further research on the maximum incremental reactivity (MIR) of methanol and ethanol in environmental chambers agrees with the findings reported above (Carter et al., 1995b). Notably the MIRs for ethanol and methanol, 1.69 and 0.71, respectively (mass ozone per mass VOC, produced under high NOx conditions), are considerably lower than average for representative ‘background’ VOCs—which have an MIR of 3.7. Although OFP for methanol and ethanol can be calculated by models or measured directly in large environmental chambers, these approaches do not account for the possibility that previously unidentified organic compounds with high OFP may be present in the emissions from dairy cows and their waste. One way to check the OFP of these unknown compounds is directly at the dairy source using realworld emissions. The purpose of this research is to directly determine the OFP of dairy cattle emissions using a mobile atmospheric chamber. The chamber is similar in design to those created and operated by previous researchers (Kamens et al., 1984; Leone et al., 1985; Carter et al., 1995a; Cocker et al., 2001; Pathak et al., 2007) but small enough to allow for easy transport to the source. The chamber was tested with a comprehensive set of quality control checks to validate its ability to accurately portray the OFP of EtOH. Field experiments were then carried out using emissions from dairy cattle in a controlled environment. Results are presented as ozone isopleths that were further validated with model calculations. 2. Methods 2.1. Chamber design Fig. 1a and b illustrates the Mobile Ozone Chamber Assay (MOChA) designed for studies of ozone formation in the field. The environmental chamber that housed the Teflon (PFA) bag and the UV light assembly were constructed using standard materials (200 400 studs and 5800 plywood). The inner surface of each chamber was coated with aluminum sheeting that had a reflectance of 95%. Both the environmental chamber and the UV light assembly were equipped with castor wheels that enabled them to be transported to difference sources. Fig. 1b shows the details of the UV light assembly that emitted the photo-chemically relevant portion of the solar spectrum in a controlled and reproducible manner. During experiments the UV light
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
5269
Fig. 1. (a) Teflon reaction chamber, environmental enclosure, sampling manifold, and (b) ultraviolet lamps used in the mobile ozone chamber assay (MOChA).
assembly was joined to the opening of the environmental chamber that housed the Teflon reaction chamber. The closest distance between the lamps and the Teflon chamber was 0.5 m. Ultra-violet light was produced using 26 F40BL Sylvania black bulbs that emit a peak intensity at a wavelength of 350 nm (Cocker et al., 2001). The overall average intensity of UV radiation in the Teflon chamber was measured with a UV photometer (Model #PMA2111, Solar Light Co. Inc., Glenside, PA) and found to be 50 W m2. Similar measurements made on a clear summer day in the SJV produced values of 55 W m2. The Teflon reaction chamber used in the MOChA system had a volume of approximately 1 m3. Larger bags could have been used but the increased bulk of the environmental chamber needed to enclose the
bag would have reduced the transportability of the system. A sampling port comprised of stainless steel tubes, Swagelok fittings, and valves was used to fill the bag before experiments and to measure concentrations in the bag during experiments. The bag was filled through manually operated ball valves using a Teflon diaphragm pump (Model #R224-BPBA1, Air Dimension Inc., Deerfield Beach, FL). Measurements of pollutant concentrations in the bag during the experiment were made via an automated three-way ball valve controlled by a sampling program. 2.2. Experimental procedure The mobile chamber was first tested under controlled conditions in the air quality laboratories
ARTICLE IN PRESS 5270
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
of the Department of Civil and Environmental Engineering at UC Davis. The chamber was filled with known concentrations of ozone (O3), nitrogen monoxide (NO), nitrogen dioxide (NO2), and different organic compounds. Background air was either generated with a zero-air generator (Model #ZA-750-12, Perma Pure Inc., Toms River, NJ) operated at 15 L min1 or background laboratory air was directly injected into the bag using the Teflon diaphragm pump (50–100 L min1). The zero-air generator has an ultra-violet lamp for the removal of microorganisms, an activated carbon/ alumina mixture for the removal of trace quantities of NOx and O3, a carbon monoxide catalyst to convert CO to CO2, and a 1 mm particulate filter with a 93% rating down to 0.1 mm. In either filling condition, a 47 mm diameter quartz fiber filter (99.9% at 0.3 mm, Pall Life Sciences, Ann Arbor, MI) was placed upstream of the bag to remove any airborne particles that may have been contained in the background air. Selected VOCs were introduced into the bag during testing by placing a measured amount of liquid VOC on the quartz filter and allowing it to volatilize into the air stream during the filling procedure. A known concentration of oxides of nitrogen (NOx) was introduced into the bag as NO2 from a highpressure cylinder, (10.670.5 ppm volume/volume NOx, balance air, Scotts Specialty Gases, Longmont, CO) with flow measured by a calibrated rotameter. O3 produced by an ozone generator (Model#1008-RS, Dasibi Environmental Corporation, Glendale, CA) was injected directly into the bag during initial experiments to characterize the wall loss rate. Once the reaction chamber was filled with the target NOx and VOC concentrations, the manual inlet valves were closed and automated sampling began. The UV lights were turned on during simulated sunlight experiments or they were left off for simulated nighttime experiments. Samples were drawn from the reaction chamber using a computercontrolled three-way ball valve over a three hour time period. The flow rate for the instrument package was 3.1 L min1, but samples were only drawn from the reaction chamber at specified intervals to prevent complete deflation of the bag during each experiment. Measurements were made from the bag from 0 to 5, 20 to 30, 55 to 65, 85 to 90, 115 to 120, 145 to 150, and 175 to 180 min. The instruments were operated continuously over the entire 3-h experiment to avoid transient data associated with startup.
During periods when instruments were not monitoring concentrations in the reaction chamber they sampled ambient air adjacent to the chamber via the 3-way valve. This sampling protocol ensured that the Teflon reaction chamber retained the majority of its initial volume during each experiment. Temperature, relative humidity (RH) and the concentration of NO, NO2, and O3 were monitored in the Teflon reaction chamber during each experiment. Concentrations of NO and NO2 were measured using a chemiluminescence analyzer (Model #ML9841A, Teledyne Monitor Labs, Englewood, CO) at a flow rate of 0.60 L min1. The NO+NO2 ( ¼ NOx) analyzer was calibrated (zero and span) using the zero-air generator and pure NO2 gas (Certified Working Class 75%, Scott Specialty Gases, Longmont, CO), respectively. The working range of the chemiluminescence analyzer was 0–20 ppm with a lower detectable limit of 0.5 ppb. Concentrations of O3 were measured using an ozone analyzer (Model #450, Teledyne Instruments Advanced Pollution Instrumentation, Inc., San Diego, CA) at a flow rate of 1.5 L min1. The ozone analyzer was initially calibrated by the manufacturer and then checked periodically against a second ozone analyzer to verify precision. The ozone analyzer had a working range of 0–1000 ppb with a lower detection limit o3 ppb). Relative humidity and temperature were measured in the sample stream at a flow rate of 1 L min1 using a relative humidity and temperature probe (Model HMP50-L, Campbell Scientific, Logan, UT). All measurements were logged using National Instruments measurement and automation software (National Instruments, Austin, TX). Light intensity was measured during each experiment with a UV photometer (Model #PMA211, Solar Light Co. Inc., Glenside, PA). Lamps were replaced when the UV intensity decreased by more than 10% relative to the initial (new product) value. Typically 50–60% of the sample air volume remained in the Teflon reaction chamber at the end of each experiment. This air was removed after each experiment using a pump with a flow rate of 20 L min1. The reaction chamber was then flushed with clean air generated by the zero-air generator at a relative humidity of 50%. All inlet and outlet ports were closed and covered with aluminum foil to decrease the possibility of contamination between experiments. The Teflon bags were regularly checked for leaks and were replaced at frequent intervals when they become dirty or unusable.
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Concentration (ppb)
The MOChA was subjected to several tests to verify proper behavior before field experiments were conducted. As a first check, the chamber was filled with clean air (no NOx or VOC) that was illuminated with the UV lamps at full intensity. Measured NOx and O3 concentrations were expected to be close to zero under these ‘‘blank’’ conditions. Fig. 2 illustrates the average results from four replicates of the blank experiment. Each point on the graph represents the average concentration over the four experiments, while the error bars show the standard deviation. The concentration of O3 and NOx was small but non-zero during the experiment because the surface of the bag is not perfectly clean leading to off-gassing during the experiment. Both NOx and O3 concentrations decreased after the initial sampling interval and then stabilized at approximately 5 ppb. The NOx measurements had an average uncertainty of 15%, while the O3 measurements had an average uncertainty of 3%. The concentrations of both NOx and O3 during the blank experiment were far lower than measured concentrations during actual experiments. Replicate experiments were performed in order to verify the stability of ozone formation under similar conditions. NOx and EtOH were injected into the bag at 40 and 1200 ppb respectively, and the formation of ozone was measured over a 180-min time period. NOx and EtOH concentrations were also measured to ensure that the concentrations for all three experiments were similar (data not shown). EtOH was measured using an INNOVA photoacoustic field gas sampler (Model 1412, INNOVA AirTech Instruments, Ballerup, Denmark). Only the experiments where NOx and EtOH were within 5% of the target values were retained in the results that Background Bag Concentrations of NOx and O3
40 35 30 25 20 15 10 5 0
Ozone Zero
0
20
40
60 80 100 Time (min)
120
NOx Zero
140
160
Fig. 2. Measured concentrations of Ozone and NOx in MOChA during blank experiments with UV lamps at full light intensity.
O3 Concentration (ppb)
2.3. Validation experiments
5271
Precision of Ozone Formation (3 Replicates)
40 35 30 25 20 15 10 5 0 0
20
40
60
80 100 120 140 160 180 Time (mins)
Fig. 3. Precision of replicate ozone formation experiments (NOx ¼ 40 ppb and EtOH ¼ 1200 ppb). Error bars illustrate the standard deviation of three experiments.
are presented in Fig. 3. The UV intensity was also measured to ensure that the experiments were performed under similar lighting conditions. The results shown in Fig. 3 demonstrate that ozone formation is stable under similar conditions within the environmental chamber. As a final check, the ozone generated by the UV lamps was compared to the ozone generated by natural sunlight (incident angle of 451 on a clear August afternoon) with resulting ozone production similar to that measured in the laboratory. The loss rate of NOx and O3 to the surface of the Teflon reaction chamber was measured separately to better understand the dynamics within each experiment. The loss-rate experiments were performed by injecting either NOx (as NO2) or ozone (from an ozone generator) into the bag and then monitoring the concentrations of the species over a 180-min time period under dark conditions. NOx and ozone were never introduced together, so the loss rate represents only bag wall interactions. Each loss rate experiment was performed three times to characterize variability. Based on the result of these experiments, the NOx and Ozone loss rates in the chamber are 15.4772.5% h1 and 14.6771.21% h1, respectively. Additional experiments were performed with NOx introduced as NO with similar results. A series of laboratory experiments were completed using a range of VOC and NOx concentrations to construct ozone formation isopleths using MOChA. Fig. 4 illustrates the ethanol–NOx ozone isopleth. The color bands represent ozone concentrations from 0 to 50 ppb. The y-axis represents initial NOx concentrations (representative of summer SJV conditions) measured in ppb, while the x-axis represents the initial ethanol concentrations
ARTICLE IN PRESS 5272
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Fig. 4. Ozone isopleths generated with NOx and EtOH using the 1 m3 MOChA system. Additional data points measured with the larger 3.6 m3 LATTE system are shown in the circles with dark borders based on their NOx and EtOH concentrations. All ozone concentrations are plotted in ppb.
measured in ppbC. The 23 individual MOChA measurements are shown as color-coded squares in Fig. 5. The overall isopleth illustrates modest OFP from ethanol, with peak concentrations reaching approximately 50 ppb in 180 min. This finding is consistent with measurements of the OFP of ethanol in the presence of a representative urban VOC mixture (Carter et al., 1995b). The results shown in Fig. 4 were repeated in a 3.6 m3 smog chamber (Larger Atmospheric Transportable Test Enclosure: LATTE) to ensure that the compact size of the 1 m3 Teflon chamber used by MOChA did not influence the results. MOChA’s compact size enables the easy transportation of the system to sources in the field, but the reduced ratio of bag surface area to volume may enhance the role of surface chemistry. LATTE uses the same type of UV lights and reflective aluminum sheeting, as well as the same design for sampling ports and measurement equipment. The reaction chamber used in LATTE was constructed from FEP Teflon, but subsequent experiments showed similar results between PFA Teflon and FEP Teflon reaction chambers. The only significant difference between LATTE and MOChA is the size of the Teflon chambers (1 vs. 3.6 m3). Six experiments were performed using LATTE under various NOx and EtOH concentrations (where the UV intensity in the LATTE was within 5% of the MOChA value). This small number of experiments is not sufficient to generate a full LATTE ozone isopleth, but the
Fig. 5. Laboratory ozone isopleths generated with NOx and EtOH using the 1 m3 MOChA system. All ozone concentrations are in ppb concentrations. Laboratory measurements are shown as squares, while actual measurements from dairy emissions are shown as circles. All data points are plotted based on their measured NOx and EtOH concentrations.
available measurements were plotted on the MOChA ozone isopleth (Fig. 4) to provide a direct comparison between the two systems. Each open circle shown in Fig. 4 indicates the LATTE ozone concentration after 180 min at the indicated NOx and EtOH concentration. The majority of the LATTE ozone measurements show good agreement with the MOChA isopleth with a correlation coefficient (R2) equal to 0.87. Variability in the ethanol concentration measurements at the start of each experiment likely accounts for the few points that did not fall within the expected range. Overall, the results of experiments conducted in the 1 and 3.6 m3 chambers are in good agreement. 2.4. Field experiments Field experiments were performed at the Department of Animal Science Cow Pen Enclosure (CPE) facility at the University of California, Davis campus during August and September of 2006. A total of eight non-lactating cows were housed in a corral pen (18.5 10 m2) with a dirt floor that was completely enclosed with a 22 11 m2 dome-like structure. The steel-framed dome construction (Legend Series Cover-all Buildings, Saskatoon, Saskatchewan, Canada) was covered with a gas impermeable cover (Intertape Polymer Group, Montreal, Quebec, Canada). The CPE has a cooling pad on the east side with air drawn through the
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
structure by two fans on the west side. Two optical sensors (Monarch Instruments, Amherst, NH) are mounted on the fans to constantly monitor rotation speed (RPM). The negative pressure mechanical ventilation (wind tunnel system) produced constant directional airflow from east to west within the CPE. Fifteen field experiments were conducted, during which background air from the SJV was drawn through the CPE and then collected in the Teflon reaction chamber (Fig. 5). Each air sample from the CPE exhaust air was drawn through a quartz fiber filter (to remove particulate matter) and into MOChA using the Teflon diaphragm pump as described in previous sections. NOx (95% NO2 and 5% NO) was added to the Teflon reaction chamber to simulate expected concentrations in a rural environment. NOx concentrations ranging from 20 to 100 ppb were studied, which is typical for rural conditions in the San Joaquin Valley (California Air Resources Board, 2008). The UV lamps were then turned on and the concentrations of O3, NO, and NO2 were monitored for 180 min. VOC concentrations were measured continuously in the CPE using the Innova Photo-acoustic Field gas monitor. The concentration of EtOH in the CPE during a bag fill was logged and this value was assumed to be the concentration in the environmental chamber. 3. Experimental results The OFP of emissions from the dairy enclosures were calculated using the D(O3–NO) method (Carter et al., 1995a). In this method, ozone formation is calculated based on O3 formation ¼ ðO3 final O3 initial Þ ðNOfinal NOinitial Þ
(1)
Fig. 5 shows the maximum OFP of emissions from CPE plotted on the ozone isopleth for EtOH–NOx measured using MOChA. Each dairy emissions experiment is plotted at the x-y coordinate determined by the measured concentration of EtOH in the dairy emissions and the concentration of NOx injected into the Teflon reaction chamber. The resulting ozone formation is shown as the color within the measured data point (open circle), which can be compared to the color of the underlying EtOH–NOx ozone isopleth. Also shown are the results from the 23 individual experiments that were
5273
performed to create the isopleth shown in Figs. 4 and 5. Experimental values are displayed as colorcoded squares at the corresponding measured EtOH–NOx concentration. The results shown in Fig. 5 demonstrate that EtOH and NOx appear to explain the majority of the ozone formation that results from cow and fresh waste emissions. All experiments conducted using the default air exchange rate in the dairy cow enclosures had EtOH concentrations o500 ppb and ozone formationo20 ppb. Experiments conducted with reduced air exchange rates produced EtOH concentrations of 1 ppm (a fivefold increase over base case) and ozone formation o35 ppb. Additional experiments were conducted at the CPE in which 50 ppb of NOx and 250 ppb of ethylene or 450 ppb of the mini-surrogate (55% ethylene, 35% n-hexane, 10% xylenes) as described by Carter (Carter et al., 1995a) was mixed with air drawn from the SJV through the CPE. The results of these experiments were compared to laboratory test using only ethylene, the mini-surrogate, or the mini-surrogate+ethanol as the VOC. In all cases, the ozone formation observed at the CPE using ethylene or the mini-surrogate is completely consistent with ethanol as the dominant VOC. 3.1. Model results In order to compare the measurements summarized in Fig. 5 with our current understanding of atmospheric chemical reactions, calculations were carried out using a modified form of the Caltech Atmospheric Chemistry Mechanism (CACM) (Griffin et al., 2002). Modifications were made to the standard CACM inputs to simulate the light intensity and the UV spectrum produced by the UV lamp assembly used with MOChA. The CACM reactions describing EtOH chemistry were expanded to accurately account for ethanol chemistry in the Teflon reaction chamber. Ethanol reacts with the hydroxyl radical according to the following equation: ETOH þ OH ! 0:11HOCH2 CH2 O2 þ 0:89CH3 CHO þ 0:89HO2
(2)
However, in the base case version of CACM, the acetylaldehyde produced by this reaction is lumped into a general aldehyde species with the properties of n-pentanal (C5 molecule). CACM also lumps the peroxy radical as a C4 species. These approximations
ARTICLE IN PRESS 5274
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
are made because EtOH chemistry is not usually dominant in large urban systems and so it makes sense to increase computational efficiency by combining species. These lumping approximations have unintended consequences when modeling simple systems dominated by EtOH, because the lumped products lead to unrealistically high amounts of ozone formation. In particular, the lumped C5 aldehyde species creates a five-carbon PAN molecule that produces ozone. The CACM lumping for EtOH was modified by explicitly representing several EtOH reactions as described by the Master Chemical Mechanism (University of Leeds, 2004). Eq. (2) above was retained in the new version of CACM. The mechanism now includes the reactions of acetylaldhyde (Eqs. (3–5), the peroxy radical (Eq. (6)), and a few of the subspecies (Eqs. (7–12))). These reactions are shown below: CH3 CHO þ OH ! CH3 CO3
(3)
CH3 CHO þ hv ! CH3 O2 þ HO2 þ CO
(4)
CH3 CHO þ NO3 ! HNO3 þ CH3 CO3
(5)
HOCH2 CH2 O2 þ NO ! NO2 þ HO2 þ 1:44HCHO þ 0:28HOCH2 CHO
(6)
HOCH2 CHO þ OH ! HOCH2 CO3 þ GLYOX þ HO2
(7) HOCH2 CHO þ hv ! HO2 þ HCHO þ HO2 þ CO (8) HOCH2 CHO þ NO3 ! HOCH2 CO3 þ HNO3 (9) GLYOX þ hv ! 3CO þ HCHO
(10)
GLYOX þ OH ! 1:2CO þ 0:6HO2 þ 0:4HCOCO3 (11) GLYOX þ NO3 ! 1:2CO þ 0:6HO2 þ HNO3 þ 0:4HCOCO3
(12)
The products of these reactions are defined explicitly in CACM, except for two species, HOCH 2CO3 and HCOCO3, which were easily lumped back into the original mechanism. Note that lumped chemical mechanisms that seek to efficiently represent ozone formation do not always conserve carbon mass for the parent VOCs or their products.
This treatment is used in many state of the art chemical mechanisms including MCM, RADM, RACM, SAPRC and CACM. The chemical equations used in the current study were taken directly from the MCM and/or CACM parent mechanisms. The revised mechanism was used to simulate O3 formation in MOChA. Fig. 6 illustrates the ability of the model to predict NOx, EtOH, and O3 concentrations during a typical laboratory experiment. Two laboratory experiments were performed using similar input conditions of NOx, EtOH, and UV intensity. The concentrations of the various species were logged every 10 min and model calculations were then performed using the same initial conditions. The solid lines with open markers represent measured average data, while the closed markers with no lines represent model data. The comparison verifies the ability of the model to accurately capture the dynamics of NOx, EtOH, and O3 over the course of a single experiment. The model predicts the expected slight increase in NO and the slight decrease in NO2 during the first 10 min of the experiment. The model also tracks ozone concentrations remarkably well over the course of the experiment. Both predicted and observed EtOH concentrations decrease by 10–15%, demonstrating ethanol’s low reactivity and low ozone forming potential with the model correctly capturing this behavior. Measured NOx concentrations increased at the end of the 180-min experiment but the model predictions continued to decrease. The observed increase in NOx concentrations may be due to off-gassing of NOx from the Teflon surface as the bag volume decreases at the end of an experiment. As mentioned previously, several lab and CPE experiments were conducted with the addition of a ‘‘mini-surrogate’’ to the usual ethanol+NOx mixture. Model simulations were also carried out for this mixture VOCs+NOx using the modified version of CACM. All model predictions for ozone concentrations were within 10% of measured values. These results demonstrate that the modified version of CACM is able to capture ozone formation dynamics for the wide range of possible VOC+NOx mixtures that could occur in the agricultural/urban environment. Fig. 7a summarizes the simulated ozone formation from dairy TOG (dominated by EtOH)–NOx mixtures at the UV intensity in the experimental chambers. Fig. 7a was created by performing 25 model calculations at NOx concentrations ranging
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Measured vs. Predicted Concentrations 35
1.20
5275
NOx Measured (ppb) NO Measured (ppb)
Concentration (ppb or ppm)
30
1.00 NO2 Measured (ppb)
25 0.80
O3 Measured (ppb)
20 0.60 15
NO Model (ppb) NO2 Model (ppb)
0.40 10
O3 Model (ppb)
0.20
5
0.00
0 0
25
50
75 100 125 150 175 200 Time (mins)
NOx Model (ppb) EtOH Measured (ppm) (2nd Axis) EtOH Model (ppm) (2nd axis)
Fig. 6. MOChA lab results (solid lines with open symbols) compared with model predictions (closed symbols). All species concentrations are expressed ppb, except for EtOH, which is plotted in ppm.
from 0 to 100 ppb and TOG concentrations ranging from 0 to 200 ppm. The speciation of detailed organic species within the TOG was based on measured concentrations (98% CH4, 1% EtOH, 0.8% MeOH) (Shaw et al., 2007). Water vapor was included in the simulation with initial concentrations equal to 50% RH at 298 K. The maximum ozone formed during each computer simulation was logged and then plotted to form the isopleth. The predicted ozone isopleth shown in Fig. 7a peaks at 50 ppb O3 when NOx is 20 ppb and TOG is 200 ppm. The actual ozone formation measured at the Dairy Cow Enclosure facility is also shown in Fig. 7a as the color-coded open circles at the equivalent TOG and NOx concentrations. Predicted and measured ozone concentrations show relatively good agreement at most TOG and NOx concentrations. Predicted ozone formation is slightly lower than actual ozone production in some cases. Some of these discrepancies could be accounted for by the VOC lumping scheme used in the model. Although the chemical mechanism in CACM was enhanced using the Master Chemical Mechanism (MCM) so that EtOH was treated as a discrete species, the model still combines a number of downstream reaction products to increase computational efficiency. Any lumping scheme is approximate, and will lead to some inaccuracy. Also, wall loss rates in the Teflon bag were approximated in the model based on measured NOx and O3 loss rates, but the
actual loss rates for these other species could vary depending on their unique chemical structure. Despite these potential shortcomings, overall agreement between predicted and measured concentrations is favorable, building confidence in both the input source profiles and the model chemical mechanism. Fig. 7b illustrates the expected ozone formation when the emission profile for dairy cattle and their waste specified by the California Air Resources Board (CARB) is used in the model calculations (California Air Resources Board, 2007). In CARB’s emission profile, CH4 accounts for only 70% of the TOG, with ethane (20%), EtOH (2%), acetone (3%) and other alcohols and amines making up the balance. The CARB profile results in a maximum ozone concentration of 140 ppb when TOG is 200 ppm and NOx is 100 ppb. Based on model sensitivity analysis (not shown), the increased ozone formation is almost entirely due to the inclusion of acetone in the CARB profile, with smaller differences attributed to ethane. 4. Discussion The SJVAPCD currently plans to decrease VOC emissions in the SJV and states that dairy cattle account for a significant fraction of ROG in this region. The experimental and model results from the current study show that the historical assumptions for the composition of organic gases emitted
ARTICLE IN PRESS 5276
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
1 m3 reaction chamber, could capture the dynamics of ozone formation in systems dominated by EtOH chemistry. Although it is virtually impossible to characterize every organic compound in a complex environmental mixture, direct measurements of OFP from dairy cows combined with previous measurements of VOCs in dairy emissions strongly suggest that virtually all of the OFP is explained by EtOH concentrations measured from dairy cows and their waste. Simulations of ozone formation from dairy sources must ensure that ozone chemistry is properly tracked without excessive consolidation of intermediate products. Simulations carried out with a modified form of the Caltech Atmospheric Chemistry Mechanism (CACM) were able to correctly predict OFP in the EtOH system and OFP from actual dairy emissions. The model calculations show that TOG profiles including acetone greatly overestimate the OFP of dairy emissions. Future studies will consider the implications of these findings for ozone formation in California. Acknowledgment This research was funded by the USDA Grant TM#2004-06138.
Fig. 7. Model predictions for ozone formation isopleths associated with dairy emissions using (a) the VOC profile dominated by EtOH (Shaw et al., 2007) and (b) using the VOC profile in the California Air Resources Board source profile library. Experimental ozone measurements are shown as circles. All ozone concentrations are ppb. Note the different scales used in panels (a) and (b).
from dairy cows and their waste significantly overestimates the OFP for this source. These initial studies suggest that decreasing ROG from dairy cattle, specifically by decreasing EtOH concentrations, will not likely lead to decreased ozone concentrations in the SJV.
5. Conclusions A mobile atmospheric chamber (MOChA) was created to directly measure the ozone formation potential (OFP) of emissions from dairy cows and their fresh waste. Comprehensive validation experiments were carried out to verify that MOChA, the
References California Air Resources Board. (2007, April 3, 2007). Dairy cattle emissions profile. Retrieved July, 2007, from /http:// www.arb.ca.gov/ei/speciate/speciate.htmS. California Air Resources Board. (2008, December 2006). San Joaquin Valley NOx concentration. Retrieved July, 2007, from /http://www.arb.ca.gov/aqd/aqdpage.htmS. Carter, W.P.L., Luo, D.M., et al., 1995a. Environmental chamber studies of atmospheric reactivities of volatile organic-compounds. Effects of varying ROG surrogate and NOx. Final Report to Coordinating Research Council, Inc., Project ME-9, California Air Resources Board, Contract A032-0692, and South Coast Air Quality Management District, Contract C91323 10 September. Carter, W.P.L., Pierce, J.A., et al., 1995b. Environmental chamber study of maximum incremental reactivities of volatile organic-compounds. Atmospheric Environment 29 (18), 2499–2511. Cocker, D.R., Flagan, R.C., et al., 2001. State-of-the-art chamber facility for studying atmospheric aerosol chemistry. Environmental Science and Technology 35 (12), 2594–2601. Department of Food and Agricultural-California. (2007). California Dairy Industry facts. Retrieved July, 2007, from /http://www.californiadairypressroom.com/CADPK_ industryfacts.htmlS.
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277 Derwent, R.G., Jenkin, M.E., et al., 1998. Photochemical ozone creation potentials for organic compounds in northwest Europe calculated with a master chemical mechanism. Atmospheric Environment 32 (14–15), 2429–2441. Griffin, R.J., Dabdub, D., et al., 2002. Secondary organic aerosol-1. Atmospheric chemical mechanism for production of molecular constituents. Journal of Geophysical Research— Atmospheres 107 (D17). Hakami, A., Harley, R.A., et al., 2004. Regional, threedimensional assessment of the ozone formation potential of organic compounds. Atmospheric Environment 38 (1), 121–134. Kamens, R.M., Rives, G.D., et al., 1984. Mutagenic changes in dilute wood smoke as it ages and reacts with ozone and nitrogen-dioxide—an outdoor chamber study. Environmental Science and Technology 18 (7), 523–530. Leone, J.A., Flagan, R.C., et al., 1985. An outdoor smog chamber and modeling study of toluene–NOx photooxidation. International Journal of Chemical Kinetics 17 (2), 177–216. Mitloehner, F.M., 2005. Effects of insufficient data on air quality regulatory policy in animal agriculture. Journal of Applied Poultry Science 14, 373–377. Pathak, R.K., Stanier, C.O., et al., 2007. Ozonolysis of alphapinene at atmospherically relevant concentrations: temperature dependence of aerosol mass fractions (yields). Journal of Geophysical Research 112 (D3).
5277
Ritzman, E.G., Benedict, F.G., 1938. Nutritional Physiology of the Adult Ruminant. Carnegie Institute, Washington, DC. Russell, A., Milford, J., et al., 1995. Urban ozone control and atmospheric reactivity of organic gases. Science 269 (5223), 491–495. San Joaquin Valley Air Pollution Control District, 2006. 1-h Extreme ozone attainment. Retrieved June 2007, from /http:// www.valleyair.org/Air_Quality_Plans/AQ_plans_Ozone_Final. htmS. San Joaquin Valley Air Pollution Control District, 2007. 8-h Ozone plan. Retrieved June, 2007, from /http://www.valleyair.org/ Air_Quality_Plans/AQ_Final_Adopted_Ozone2007.htmS. Shaw, S.L., Mitloehner, F.M., et al., 2007. Volatile organic compound emissions from dairy cows and their waste as measured by proton-transfer-reaction mass spectrometry. Environmental Science and Technology 41 (4), 1310–1316. University of Leeds, 2004. Master Chemical Mechanism. Version 3.1. Retrieved March, 2007, from /http://mcm.leeds.ac.uk/ MCM/S. U.S. Environmental Protection Agency, 1980. Volatile Organic Compound (VOC) Species Data Manual second ed.: EPA450/4-80.-015. U.S. Environmental Protection Agency, 2007. Proposed revisions to ozone national standards. EPA-HQ-OAR-2005-0172. Retrieved July, 2007, from /http://www.epa.gov/groundlevelozone/pdfs/ 20070620_fs.pdfS.
Atmospheric Environment 42 (2008) 5267–5277 www.elsevier.com/locate/atmosenv
Direct measurements of the ozone formation potential from dairy cattle emissions using a transportable smog chamber Cody J. Howarda, Wenli Yangb, Peter G. Greena, Frank Mitloehnerc, Irina L. Malkinac, Robert G. Flocchinib, Michael J. Kleemana, a
Department of Civil and Environmental Engineering, University of California at Davis, 1 Shields Avenue, Davis, CA 95616, USA b Crocker Nuclear Laboratory, University of California at Davis, 1 Shields Avenue, Davis, CA 95616, USA c Department of Animal Science, University of California at Davis, 1 Shields Avenue, Davis, CA 95616, USA Received 15 August 2007; received in revised form 8 February 2008; accepted 28 February 2008
Abstract Tropospheric ozone continues to be an air pollution problem in the United States, particularly in California, Texas, and across the eastern seaboard. The obvious sources of ozone precursors have been largely controlled over the past several decades, leading to the critical examination of secondary sources. In particular, California has new air quality rules addressing agricultural sources of ozone precursors, including dairy farms. Some recent estimates predict that dairy cattle are second only to on-road vehicles as a leading source of ozone precursor emissions in California’s San Joaquin Valley. The objective of this work was to directly measure the ozone formation potential from dairy housing. A transportable ‘‘smog’’ chamber was constructed and validated using organic gases known to be present in dairy emissions. The ozone formation potential of emissions from eight non-lactating dairy cows and their fresh waste was then directly evaluated in the field at a completely enclosed cow corral on the campus of the University of California, Davis. The results demonstrate that the majority of the ozone formation is explained by ethanol (EtOH) in the emissions from the dairy cows, not by acetone as previously thought. Ozone formation potential is generally small, with o20 ppb of ozone produced under typical conditions when EtOH concentrations were 200 ppb and NOx concentrations were 50 ppb. The results match our current understanding of atmospheric ozone formation potential, ruling out the possibility of unknown organic compounds in dairy emissions with significant ozone formation potential. Simulations carried out with a modified form of the Caltech Atmospheric Chemistry Mechanism verify that actual ozone formation from dairy emissions is much lower than what would be predicted using the current regulatory profiles. Based on these results, the ozone formation potential of emissions from dairy cattle in California seems to be lower than previously estimated. r 2008 Elsevier Ltd. All rights reserved. Keywords: Ozone formation potential; Portable smog chamber; Ethanol; Acetone
1. Introduction
Corresponding author. Tel.: +1 530 752 8386;
fax: +1 530 752 7872. E-mail address: [email protected] (M.J. Kleeman). 1352-2310/$ - see front matter r 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2008.02.064
Ground level ozone continues to be a problem throughout the United States (US Environmental Protection Agency, 2007) and particularly in California’s San Joaquin Valley (SJV) and South Coast
ARTICLE IN PRESS 5268
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Air Basin (SoCAB). Although mobile sources still account for the large majority of reactive organic gases (ROG) (photochemically reactive and not exempt as an ozone precursor), California has made great strides in decreasing those emissions and has begun looking for other sources, which could lead to ozone formation. One such source that has recently come under scrutiny is livestock waste, specifically from dairy cattle. Dairy is the dominant livestock industry in California, where 1.78 million dairy cows produced 38.8 billion pounds of milk in 2006, generating 21% of the national supply (Department of Food and Agricultural-California, 2007). According to the recently approved ozone plan for the SJV, livestock waste from dairy cattle account for 17.6 ton of volatile organic compounds (VOC, organic compounds which readily evaporate to the atmosphere) emissions per day and 16% of the ROG in the SJV (San Joaquin Valley Air Pollution Control District, 2007). Only mobile sources contributed more to the total ROG emitted in the region. Based on these data, the San Joaquin Valley Air Pollution Control District (SJVAPCD) approved a plan which aims to decrease VOC emission by 15.8 ton day1 from all concentrated animal feeding operations (CAFOs), with dairy farms accounting for 80% of all CAFOs (San Joaquin Valley Air Pollution Control District, 2006). The SJVAPCD based their estimates for the ozone forming potential (OFP) of dairy cow emissions on research performed on total organic gas (TOG) emissions from cows in the 1930s (Ritzman and Benedict, 1938) and an erroneously determined ROG/TOG factor (US Environmental Protection Agency, 1980; Mitloehner, 2005). Recent findings suggest that ROG emissions from dairy farms may be much lower than previously speculated and the potential for ozone formation from these sources may therefore be less than anticipated. According to Shaw et al., (2007), ROG fluxes from dairy cattle are approximately 6–10 times lower than the value suggested by present emission estimates. They calculated an ozone formation potential (OFP) of dairy cattle emissions that is 10% lower than that from combustion and biogenic sources (Shaw et al., 2007). They base this result on three separate OFP model calculations of ethanol and methanol (Russell et al., 1995; Derwent et al., 1998; Hakami et al., 2004), which are reported to account for 80–90% of the ROG in the dairy emissions and make up 1.8% of the TOG emissions
(Shaw et al., 2007). Further research on the maximum incremental reactivity (MIR) of methanol and ethanol in environmental chambers agrees with the findings reported above (Carter et al., 1995b). Notably the MIRs for ethanol and methanol, 1.69 and 0.71, respectively (mass ozone per mass VOC, produced under high NOx conditions), are considerably lower than average for representative ‘background’ VOCs—which have an MIR of 3.7. Although OFP for methanol and ethanol can be calculated by models or measured directly in large environmental chambers, these approaches do not account for the possibility that previously unidentified organic compounds with high OFP may be present in the emissions from dairy cows and their waste. One way to check the OFP of these unknown compounds is directly at the dairy source using realworld emissions. The purpose of this research is to directly determine the OFP of dairy cattle emissions using a mobile atmospheric chamber. The chamber is similar in design to those created and operated by previous researchers (Kamens et al., 1984; Leone et al., 1985; Carter et al., 1995a; Cocker et al., 2001; Pathak et al., 2007) but small enough to allow for easy transport to the source. The chamber was tested with a comprehensive set of quality control checks to validate its ability to accurately portray the OFP of EtOH. Field experiments were then carried out using emissions from dairy cattle in a controlled environment. Results are presented as ozone isopleths that were further validated with model calculations. 2. Methods 2.1. Chamber design Fig. 1a and b illustrates the Mobile Ozone Chamber Assay (MOChA) designed for studies of ozone formation in the field. The environmental chamber that housed the Teflon (PFA) bag and the UV light assembly were constructed using standard materials (200 400 studs and 5800 plywood). The inner surface of each chamber was coated with aluminum sheeting that had a reflectance of 95%. Both the environmental chamber and the UV light assembly were equipped with castor wheels that enabled them to be transported to difference sources. Fig. 1b shows the details of the UV light assembly that emitted the photo-chemically relevant portion of the solar spectrum in a controlled and reproducible manner. During experiments the UV light
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
5269
Fig. 1. (a) Teflon reaction chamber, environmental enclosure, sampling manifold, and (b) ultraviolet lamps used in the mobile ozone chamber assay (MOChA).
assembly was joined to the opening of the environmental chamber that housed the Teflon reaction chamber. The closest distance between the lamps and the Teflon chamber was 0.5 m. Ultra-violet light was produced using 26 F40BL Sylvania black bulbs that emit a peak intensity at a wavelength of 350 nm (Cocker et al., 2001). The overall average intensity of UV radiation in the Teflon chamber was measured with a UV photometer (Model #PMA2111, Solar Light Co. Inc., Glenside, PA) and found to be 50 W m2. Similar measurements made on a clear summer day in the SJV produced values of 55 W m2. The Teflon reaction chamber used in the MOChA system had a volume of approximately 1 m3. Larger bags could have been used but the increased bulk of the environmental chamber needed to enclose the
bag would have reduced the transportability of the system. A sampling port comprised of stainless steel tubes, Swagelok fittings, and valves was used to fill the bag before experiments and to measure concentrations in the bag during experiments. The bag was filled through manually operated ball valves using a Teflon diaphragm pump (Model #R224-BPBA1, Air Dimension Inc., Deerfield Beach, FL). Measurements of pollutant concentrations in the bag during the experiment were made via an automated three-way ball valve controlled by a sampling program. 2.2. Experimental procedure The mobile chamber was first tested under controlled conditions in the air quality laboratories
ARTICLE IN PRESS 5270
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
of the Department of Civil and Environmental Engineering at UC Davis. The chamber was filled with known concentrations of ozone (O3), nitrogen monoxide (NO), nitrogen dioxide (NO2), and different organic compounds. Background air was either generated with a zero-air generator (Model #ZA-750-12, Perma Pure Inc., Toms River, NJ) operated at 15 L min1 or background laboratory air was directly injected into the bag using the Teflon diaphragm pump (50–100 L min1). The zero-air generator has an ultra-violet lamp for the removal of microorganisms, an activated carbon/ alumina mixture for the removal of trace quantities of NOx and O3, a carbon monoxide catalyst to convert CO to CO2, and a 1 mm particulate filter with a 93% rating down to 0.1 mm. In either filling condition, a 47 mm diameter quartz fiber filter (99.9% at 0.3 mm, Pall Life Sciences, Ann Arbor, MI) was placed upstream of the bag to remove any airborne particles that may have been contained in the background air. Selected VOCs were introduced into the bag during testing by placing a measured amount of liquid VOC on the quartz filter and allowing it to volatilize into the air stream during the filling procedure. A known concentration of oxides of nitrogen (NOx) was introduced into the bag as NO2 from a highpressure cylinder, (10.670.5 ppm volume/volume NOx, balance air, Scotts Specialty Gases, Longmont, CO) with flow measured by a calibrated rotameter. O3 produced by an ozone generator (Model#1008-RS, Dasibi Environmental Corporation, Glendale, CA) was injected directly into the bag during initial experiments to characterize the wall loss rate. Once the reaction chamber was filled with the target NOx and VOC concentrations, the manual inlet valves were closed and automated sampling began. The UV lights were turned on during simulated sunlight experiments or they were left off for simulated nighttime experiments. Samples were drawn from the reaction chamber using a computercontrolled three-way ball valve over a three hour time period. The flow rate for the instrument package was 3.1 L min1, but samples were only drawn from the reaction chamber at specified intervals to prevent complete deflation of the bag during each experiment. Measurements were made from the bag from 0 to 5, 20 to 30, 55 to 65, 85 to 90, 115 to 120, 145 to 150, and 175 to 180 min. The instruments were operated continuously over the entire 3-h experiment to avoid transient data associated with startup.
During periods when instruments were not monitoring concentrations in the reaction chamber they sampled ambient air adjacent to the chamber via the 3-way valve. This sampling protocol ensured that the Teflon reaction chamber retained the majority of its initial volume during each experiment. Temperature, relative humidity (RH) and the concentration of NO, NO2, and O3 were monitored in the Teflon reaction chamber during each experiment. Concentrations of NO and NO2 were measured using a chemiluminescence analyzer (Model #ML9841A, Teledyne Monitor Labs, Englewood, CO) at a flow rate of 0.60 L min1. The NO+NO2 ( ¼ NOx) analyzer was calibrated (zero and span) using the zero-air generator and pure NO2 gas (Certified Working Class 75%, Scott Specialty Gases, Longmont, CO), respectively. The working range of the chemiluminescence analyzer was 0–20 ppm with a lower detectable limit of 0.5 ppb. Concentrations of O3 were measured using an ozone analyzer (Model #450, Teledyne Instruments Advanced Pollution Instrumentation, Inc., San Diego, CA) at a flow rate of 1.5 L min1. The ozone analyzer was initially calibrated by the manufacturer and then checked periodically against a second ozone analyzer to verify precision. The ozone analyzer had a working range of 0–1000 ppb with a lower detection limit o3 ppb). Relative humidity and temperature were measured in the sample stream at a flow rate of 1 L min1 using a relative humidity and temperature probe (Model HMP50-L, Campbell Scientific, Logan, UT). All measurements were logged using National Instruments measurement and automation software (National Instruments, Austin, TX). Light intensity was measured during each experiment with a UV photometer (Model #PMA211, Solar Light Co. Inc., Glenside, PA). Lamps were replaced when the UV intensity decreased by more than 10% relative to the initial (new product) value. Typically 50–60% of the sample air volume remained in the Teflon reaction chamber at the end of each experiment. This air was removed after each experiment using a pump with a flow rate of 20 L min1. The reaction chamber was then flushed with clean air generated by the zero-air generator at a relative humidity of 50%. All inlet and outlet ports were closed and covered with aluminum foil to decrease the possibility of contamination between experiments. The Teflon bags were regularly checked for leaks and were replaced at frequent intervals when they become dirty or unusable.
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Concentration (ppb)
The MOChA was subjected to several tests to verify proper behavior before field experiments were conducted. As a first check, the chamber was filled with clean air (no NOx or VOC) that was illuminated with the UV lamps at full intensity. Measured NOx and O3 concentrations were expected to be close to zero under these ‘‘blank’’ conditions. Fig. 2 illustrates the average results from four replicates of the blank experiment. Each point on the graph represents the average concentration over the four experiments, while the error bars show the standard deviation. The concentration of O3 and NOx was small but non-zero during the experiment because the surface of the bag is not perfectly clean leading to off-gassing during the experiment. Both NOx and O3 concentrations decreased after the initial sampling interval and then stabilized at approximately 5 ppb. The NOx measurements had an average uncertainty of 15%, while the O3 measurements had an average uncertainty of 3%. The concentrations of both NOx and O3 during the blank experiment were far lower than measured concentrations during actual experiments. Replicate experiments were performed in order to verify the stability of ozone formation under similar conditions. NOx and EtOH were injected into the bag at 40 and 1200 ppb respectively, and the formation of ozone was measured over a 180-min time period. NOx and EtOH concentrations were also measured to ensure that the concentrations for all three experiments were similar (data not shown). EtOH was measured using an INNOVA photoacoustic field gas sampler (Model 1412, INNOVA AirTech Instruments, Ballerup, Denmark). Only the experiments where NOx and EtOH were within 5% of the target values were retained in the results that Background Bag Concentrations of NOx and O3
40 35 30 25 20 15 10 5 0
Ozone Zero
0
20
40
60 80 100 Time (min)
120
NOx Zero
140
160
Fig. 2. Measured concentrations of Ozone and NOx in MOChA during blank experiments with UV lamps at full light intensity.
O3 Concentration (ppb)
2.3. Validation experiments
5271
Precision of Ozone Formation (3 Replicates)
40 35 30 25 20 15 10 5 0 0
20
40
60
80 100 120 140 160 180 Time (mins)
Fig. 3. Precision of replicate ozone formation experiments (NOx ¼ 40 ppb and EtOH ¼ 1200 ppb). Error bars illustrate the standard deviation of three experiments.
are presented in Fig. 3. The UV intensity was also measured to ensure that the experiments were performed under similar lighting conditions. The results shown in Fig. 3 demonstrate that ozone formation is stable under similar conditions within the environmental chamber. As a final check, the ozone generated by the UV lamps was compared to the ozone generated by natural sunlight (incident angle of 451 on a clear August afternoon) with resulting ozone production similar to that measured in the laboratory. The loss rate of NOx and O3 to the surface of the Teflon reaction chamber was measured separately to better understand the dynamics within each experiment. The loss-rate experiments were performed by injecting either NOx (as NO2) or ozone (from an ozone generator) into the bag and then monitoring the concentrations of the species over a 180-min time period under dark conditions. NOx and ozone were never introduced together, so the loss rate represents only bag wall interactions. Each loss rate experiment was performed three times to characterize variability. Based on the result of these experiments, the NOx and Ozone loss rates in the chamber are 15.4772.5% h1 and 14.6771.21% h1, respectively. Additional experiments were performed with NOx introduced as NO with similar results. A series of laboratory experiments were completed using a range of VOC and NOx concentrations to construct ozone formation isopleths using MOChA. Fig. 4 illustrates the ethanol–NOx ozone isopleth. The color bands represent ozone concentrations from 0 to 50 ppb. The y-axis represents initial NOx concentrations (representative of summer SJV conditions) measured in ppb, while the x-axis represents the initial ethanol concentrations
ARTICLE IN PRESS 5272
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Fig. 4. Ozone isopleths generated with NOx and EtOH using the 1 m3 MOChA system. Additional data points measured with the larger 3.6 m3 LATTE system are shown in the circles with dark borders based on their NOx and EtOH concentrations. All ozone concentrations are plotted in ppb.
measured in ppbC. The 23 individual MOChA measurements are shown as color-coded squares in Fig. 5. The overall isopleth illustrates modest OFP from ethanol, with peak concentrations reaching approximately 50 ppb in 180 min. This finding is consistent with measurements of the OFP of ethanol in the presence of a representative urban VOC mixture (Carter et al., 1995b). The results shown in Fig. 4 were repeated in a 3.6 m3 smog chamber (Larger Atmospheric Transportable Test Enclosure: LATTE) to ensure that the compact size of the 1 m3 Teflon chamber used by MOChA did not influence the results. MOChA’s compact size enables the easy transportation of the system to sources in the field, but the reduced ratio of bag surface area to volume may enhance the role of surface chemistry. LATTE uses the same type of UV lights and reflective aluminum sheeting, as well as the same design for sampling ports and measurement equipment. The reaction chamber used in LATTE was constructed from FEP Teflon, but subsequent experiments showed similar results between PFA Teflon and FEP Teflon reaction chambers. The only significant difference between LATTE and MOChA is the size of the Teflon chambers (1 vs. 3.6 m3). Six experiments were performed using LATTE under various NOx and EtOH concentrations (where the UV intensity in the LATTE was within 5% of the MOChA value). This small number of experiments is not sufficient to generate a full LATTE ozone isopleth, but the
Fig. 5. Laboratory ozone isopleths generated with NOx and EtOH using the 1 m3 MOChA system. All ozone concentrations are in ppb concentrations. Laboratory measurements are shown as squares, while actual measurements from dairy emissions are shown as circles. All data points are plotted based on their measured NOx and EtOH concentrations.
available measurements were plotted on the MOChA ozone isopleth (Fig. 4) to provide a direct comparison between the two systems. Each open circle shown in Fig. 4 indicates the LATTE ozone concentration after 180 min at the indicated NOx and EtOH concentration. The majority of the LATTE ozone measurements show good agreement with the MOChA isopleth with a correlation coefficient (R2) equal to 0.87. Variability in the ethanol concentration measurements at the start of each experiment likely accounts for the few points that did not fall within the expected range. Overall, the results of experiments conducted in the 1 and 3.6 m3 chambers are in good agreement. 2.4. Field experiments Field experiments were performed at the Department of Animal Science Cow Pen Enclosure (CPE) facility at the University of California, Davis campus during August and September of 2006. A total of eight non-lactating cows were housed in a corral pen (18.5 10 m2) with a dirt floor that was completely enclosed with a 22 11 m2 dome-like structure. The steel-framed dome construction (Legend Series Cover-all Buildings, Saskatoon, Saskatchewan, Canada) was covered with a gas impermeable cover (Intertape Polymer Group, Montreal, Quebec, Canada). The CPE has a cooling pad on the east side with air drawn through the
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
structure by two fans on the west side. Two optical sensors (Monarch Instruments, Amherst, NH) are mounted on the fans to constantly monitor rotation speed (RPM). The negative pressure mechanical ventilation (wind tunnel system) produced constant directional airflow from east to west within the CPE. Fifteen field experiments were conducted, during which background air from the SJV was drawn through the CPE and then collected in the Teflon reaction chamber (Fig. 5). Each air sample from the CPE exhaust air was drawn through a quartz fiber filter (to remove particulate matter) and into MOChA using the Teflon diaphragm pump as described in previous sections. NOx (95% NO2 and 5% NO) was added to the Teflon reaction chamber to simulate expected concentrations in a rural environment. NOx concentrations ranging from 20 to 100 ppb were studied, which is typical for rural conditions in the San Joaquin Valley (California Air Resources Board, 2008). The UV lamps were then turned on and the concentrations of O3, NO, and NO2 were monitored for 180 min. VOC concentrations were measured continuously in the CPE using the Innova Photo-acoustic Field gas monitor. The concentration of EtOH in the CPE during a bag fill was logged and this value was assumed to be the concentration in the environmental chamber. 3. Experimental results The OFP of emissions from the dairy enclosures were calculated using the D(O3–NO) method (Carter et al., 1995a). In this method, ozone formation is calculated based on O3 formation ¼ ðO3 final O3 initial Þ ðNOfinal NOinitial Þ
(1)
Fig. 5 shows the maximum OFP of emissions from CPE plotted on the ozone isopleth for EtOH–NOx measured using MOChA. Each dairy emissions experiment is plotted at the x-y coordinate determined by the measured concentration of EtOH in the dairy emissions and the concentration of NOx injected into the Teflon reaction chamber. The resulting ozone formation is shown as the color within the measured data point (open circle), which can be compared to the color of the underlying EtOH–NOx ozone isopleth. Also shown are the results from the 23 individual experiments that were
5273
performed to create the isopleth shown in Figs. 4 and 5. Experimental values are displayed as colorcoded squares at the corresponding measured EtOH–NOx concentration. The results shown in Fig. 5 demonstrate that EtOH and NOx appear to explain the majority of the ozone formation that results from cow and fresh waste emissions. All experiments conducted using the default air exchange rate in the dairy cow enclosures had EtOH concentrations o500 ppb and ozone formationo20 ppb. Experiments conducted with reduced air exchange rates produced EtOH concentrations of 1 ppm (a fivefold increase over base case) and ozone formation o35 ppb. Additional experiments were conducted at the CPE in which 50 ppb of NOx and 250 ppb of ethylene or 450 ppb of the mini-surrogate (55% ethylene, 35% n-hexane, 10% xylenes) as described by Carter (Carter et al., 1995a) was mixed with air drawn from the SJV through the CPE. The results of these experiments were compared to laboratory test using only ethylene, the mini-surrogate, or the mini-surrogate+ethanol as the VOC. In all cases, the ozone formation observed at the CPE using ethylene or the mini-surrogate is completely consistent with ethanol as the dominant VOC. 3.1. Model results In order to compare the measurements summarized in Fig. 5 with our current understanding of atmospheric chemical reactions, calculations were carried out using a modified form of the Caltech Atmospheric Chemistry Mechanism (CACM) (Griffin et al., 2002). Modifications were made to the standard CACM inputs to simulate the light intensity and the UV spectrum produced by the UV lamp assembly used with MOChA. The CACM reactions describing EtOH chemistry were expanded to accurately account for ethanol chemistry in the Teflon reaction chamber. Ethanol reacts with the hydroxyl radical according to the following equation: ETOH þ OH ! 0:11HOCH2 CH2 O2 þ 0:89CH3 CHO þ 0:89HO2
(2)
However, in the base case version of CACM, the acetylaldehyde produced by this reaction is lumped into a general aldehyde species with the properties of n-pentanal (C5 molecule). CACM also lumps the peroxy radical as a C4 species. These approximations
ARTICLE IN PRESS 5274
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
are made because EtOH chemistry is not usually dominant in large urban systems and so it makes sense to increase computational efficiency by combining species. These lumping approximations have unintended consequences when modeling simple systems dominated by EtOH, because the lumped products lead to unrealistically high amounts of ozone formation. In particular, the lumped C5 aldehyde species creates a five-carbon PAN molecule that produces ozone. The CACM lumping for EtOH was modified by explicitly representing several EtOH reactions as described by the Master Chemical Mechanism (University of Leeds, 2004). Eq. (2) above was retained in the new version of CACM. The mechanism now includes the reactions of acetylaldhyde (Eqs. (3–5), the peroxy radical (Eq. (6)), and a few of the subspecies (Eqs. (7–12))). These reactions are shown below: CH3 CHO þ OH ! CH3 CO3
(3)
CH3 CHO þ hv ! CH3 O2 þ HO2 þ CO
(4)
CH3 CHO þ NO3 ! HNO3 þ CH3 CO3
(5)
HOCH2 CH2 O2 þ NO ! NO2 þ HO2 þ 1:44HCHO þ 0:28HOCH2 CHO
(6)
HOCH2 CHO þ OH ! HOCH2 CO3 þ GLYOX þ HO2
(7) HOCH2 CHO þ hv ! HO2 þ HCHO þ HO2 þ CO (8) HOCH2 CHO þ NO3 ! HOCH2 CO3 þ HNO3 (9) GLYOX þ hv ! 3CO þ HCHO
(10)
GLYOX þ OH ! 1:2CO þ 0:6HO2 þ 0:4HCOCO3 (11) GLYOX þ NO3 ! 1:2CO þ 0:6HO2 þ HNO3 þ 0:4HCOCO3
(12)
The products of these reactions are defined explicitly in CACM, except for two species, HOCH 2CO3 and HCOCO3, which were easily lumped back into the original mechanism. Note that lumped chemical mechanisms that seek to efficiently represent ozone formation do not always conserve carbon mass for the parent VOCs or their products.
This treatment is used in many state of the art chemical mechanisms including MCM, RADM, RACM, SAPRC and CACM. The chemical equations used in the current study were taken directly from the MCM and/or CACM parent mechanisms. The revised mechanism was used to simulate O3 formation in MOChA. Fig. 6 illustrates the ability of the model to predict NOx, EtOH, and O3 concentrations during a typical laboratory experiment. Two laboratory experiments were performed using similar input conditions of NOx, EtOH, and UV intensity. The concentrations of the various species were logged every 10 min and model calculations were then performed using the same initial conditions. The solid lines with open markers represent measured average data, while the closed markers with no lines represent model data. The comparison verifies the ability of the model to accurately capture the dynamics of NOx, EtOH, and O3 over the course of a single experiment. The model predicts the expected slight increase in NO and the slight decrease in NO2 during the first 10 min of the experiment. The model also tracks ozone concentrations remarkably well over the course of the experiment. Both predicted and observed EtOH concentrations decrease by 10–15%, demonstrating ethanol’s low reactivity and low ozone forming potential with the model correctly capturing this behavior. Measured NOx concentrations increased at the end of the 180-min experiment but the model predictions continued to decrease. The observed increase in NOx concentrations may be due to off-gassing of NOx from the Teflon surface as the bag volume decreases at the end of an experiment. As mentioned previously, several lab and CPE experiments were conducted with the addition of a ‘‘mini-surrogate’’ to the usual ethanol+NOx mixture. Model simulations were also carried out for this mixture VOCs+NOx using the modified version of CACM. All model predictions for ozone concentrations were within 10% of measured values. These results demonstrate that the modified version of CACM is able to capture ozone formation dynamics for the wide range of possible VOC+NOx mixtures that could occur in the agricultural/urban environment. Fig. 7a summarizes the simulated ozone formation from dairy TOG (dominated by EtOH)–NOx mixtures at the UV intensity in the experimental chambers. Fig. 7a was created by performing 25 model calculations at NOx concentrations ranging
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
Measured vs. Predicted Concentrations 35
1.20
5275
NOx Measured (ppb) NO Measured (ppb)
Concentration (ppb or ppm)
30
1.00 NO2 Measured (ppb)
25 0.80
O3 Measured (ppb)
20 0.60 15
NO Model (ppb) NO2 Model (ppb)
0.40 10
O3 Model (ppb)
0.20
5
0.00
0 0
25
50
75 100 125 150 175 200 Time (mins)
NOx Model (ppb) EtOH Measured (ppm) (2nd Axis) EtOH Model (ppm) (2nd axis)
Fig. 6. MOChA lab results (solid lines with open symbols) compared with model predictions (closed symbols). All species concentrations are expressed ppb, except for EtOH, which is plotted in ppm.
from 0 to 100 ppb and TOG concentrations ranging from 0 to 200 ppm. The speciation of detailed organic species within the TOG was based on measured concentrations (98% CH4, 1% EtOH, 0.8% MeOH) (Shaw et al., 2007). Water vapor was included in the simulation with initial concentrations equal to 50% RH at 298 K. The maximum ozone formed during each computer simulation was logged and then plotted to form the isopleth. The predicted ozone isopleth shown in Fig. 7a peaks at 50 ppb O3 when NOx is 20 ppb and TOG is 200 ppm. The actual ozone formation measured at the Dairy Cow Enclosure facility is also shown in Fig. 7a as the color-coded open circles at the equivalent TOG and NOx concentrations. Predicted and measured ozone concentrations show relatively good agreement at most TOG and NOx concentrations. Predicted ozone formation is slightly lower than actual ozone production in some cases. Some of these discrepancies could be accounted for by the VOC lumping scheme used in the model. Although the chemical mechanism in CACM was enhanced using the Master Chemical Mechanism (MCM) so that EtOH was treated as a discrete species, the model still combines a number of downstream reaction products to increase computational efficiency. Any lumping scheme is approximate, and will lead to some inaccuracy. Also, wall loss rates in the Teflon bag were approximated in the model based on measured NOx and O3 loss rates, but the
actual loss rates for these other species could vary depending on their unique chemical structure. Despite these potential shortcomings, overall agreement between predicted and measured concentrations is favorable, building confidence in both the input source profiles and the model chemical mechanism. Fig. 7b illustrates the expected ozone formation when the emission profile for dairy cattle and their waste specified by the California Air Resources Board (CARB) is used in the model calculations (California Air Resources Board, 2007). In CARB’s emission profile, CH4 accounts for only 70% of the TOG, with ethane (20%), EtOH (2%), acetone (3%) and other alcohols and amines making up the balance. The CARB profile results in a maximum ozone concentration of 140 ppb when TOG is 200 ppm and NOx is 100 ppb. Based on model sensitivity analysis (not shown), the increased ozone formation is almost entirely due to the inclusion of acetone in the CARB profile, with smaller differences attributed to ethane. 4. Discussion The SJVAPCD currently plans to decrease VOC emissions in the SJV and states that dairy cattle account for a significant fraction of ROG in this region. The experimental and model results from the current study show that the historical assumptions for the composition of organic gases emitted
ARTICLE IN PRESS 5276
C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277
1 m3 reaction chamber, could capture the dynamics of ozone formation in systems dominated by EtOH chemistry. Although it is virtually impossible to characterize every organic compound in a complex environmental mixture, direct measurements of OFP from dairy cows combined with previous measurements of VOCs in dairy emissions strongly suggest that virtually all of the OFP is explained by EtOH concentrations measured from dairy cows and their waste. Simulations of ozone formation from dairy sources must ensure that ozone chemistry is properly tracked without excessive consolidation of intermediate products. Simulations carried out with a modified form of the Caltech Atmospheric Chemistry Mechanism (CACM) were able to correctly predict OFP in the EtOH system and OFP from actual dairy emissions. The model calculations show that TOG profiles including acetone greatly overestimate the OFP of dairy emissions. Future studies will consider the implications of these findings for ozone formation in California. Acknowledgment This research was funded by the USDA Grant TM#2004-06138.
Fig. 7. Model predictions for ozone formation isopleths associated with dairy emissions using (a) the VOC profile dominated by EtOH (Shaw et al., 2007) and (b) using the VOC profile in the California Air Resources Board source profile library. Experimental ozone measurements are shown as circles. All ozone concentrations are ppb. Note the different scales used in panels (a) and (b).
from dairy cows and their waste significantly overestimates the OFP for this source. These initial studies suggest that decreasing ROG from dairy cattle, specifically by decreasing EtOH concentrations, will not likely lead to decreased ozone concentrations in the SJV.
5. Conclusions A mobile atmospheric chamber (MOChA) was created to directly measure the ozone formation potential (OFP) of emissions from dairy cows and their fresh waste. Comprehensive validation experiments were carried out to verify that MOChA, the
References California Air Resources Board. (2007, April 3, 2007). Dairy cattle emissions profile. Retrieved July, 2007, from /http:// www.arb.ca.gov/ei/speciate/speciate.htmS. California Air Resources Board. (2008, December 2006). San Joaquin Valley NOx concentration. Retrieved July, 2007, from /http://www.arb.ca.gov/aqd/aqdpage.htmS. Carter, W.P.L., Luo, D.M., et al., 1995a. Environmental chamber studies of atmospheric reactivities of volatile organic-compounds. Effects of varying ROG surrogate and NOx. Final Report to Coordinating Research Council, Inc., Project ME-9, California Air Resources Board, Contract A032-0692, and South Coast Air Quality Management District, Contract C91323 10 September. Carter, W.P.L., Pierce, J.A., et al., 1995b. Environmental chamber study of maximum incremental reactivities of volatile organic-compounds. Atmospheric Environment 29 (18), 2499–2511. Cocker, D.R., Flagan, R.C., et al., 2001. State-of-the-art chamber facility for studying atmospheric aerosol chemistry. Environmental Science and Technology 35 (12), 2594–2601. Department of Food and Agricultural-California. (2007). California Dairy Industry facts. Retrieved July, 2007, from /http://www.californiadairypressroom.com/CADPK_ industryfacts.htmlS.
ARTICLE IN PRESS C.J. Howard et al. / Atmospheric Environment 42 (2008) 5267–5277 Derwent, R.G., Jenkin, M.E., et al., 1998. Photochemical ozone creation potentials for organic compounds in northwest Europe calculated with a master chemical mechanism. Atmospheric Environment 32 (14–15), 2429–2441. Griffin, R.J., Dabdub, D., et al., 2002. Secondary organic aerosol-1. Atmospheric chemical mechanism for production of molecular constituents. Journal of Geophysical Research— Atmospheres 107 (D17). Hakami, A., Harley, R.A., et al., 2004. Regional, threedimensional assessment of the ozone formation potential of organic compounds. Atmospheric Environment 38 (1), 121–134. Kamens, R.M., Rives, G.D., et al., 1984. Mutagenic changes in dilute wood smoke as it ages and reacts with ozone and nitrogen-dioxide—an outdoor chamber study. Environmental Science and Technology 18 (7), 523–530. Leone, J.A., Flagan, R.C., et al., 1985. An outdoor smog chamber and modeling study of toluene–NOx photooxidation. International Journal of Chemical Kinetics 17 (2), 177–216. Mitloehner, F.M., 2005. Effects of insufficient data on air quality regulatory policy in animal agriculture. Journal of Applied Poultry Science 14, 373–377. Pathak, R.K., Stanier, C.O., et al., 2007. Ozonolysis of alphapinene at atmospherically relevant concentrations: temperature dependence of aerosol mass fractions (yields). Journal of Geophysical Research 112 (D3).
5277
Ritzman, E.G., Benedict, F.G., 1938. Nutritional Physiology of the Adult Ruminant. Carnegie Institute, Washington, DC. Russell, A., Milford, J., et al., 1995. Urban ozone control and atmospheric reactivity of organic gases. Science 269 (5223), 491–495. San Joaquin Valley Air Pollution Control District, 2006. 1-h Extreme ozone attainment. Retrieved June 2007, from /http:// www.valleyair.org/Air_Quality_Plans/AQ_plans_Ozone_Final. htmS. San Joaquin Valley Air Pollution Control District, 2007. 8-h Ozone plan. Retrieved June, 2007, from /http://www.valleyair.org/ Air_Quality_Plans/AQ_Final_Adopted_Ozone2007.htmS. Shaw, S.L., Mitloehner, F.M., et al., 2007. Volatile organic compound emissions from dairy cows and their waste as measured by proton-transfer-reaction mass spectrometry. Environmental Science and Technology 41 (4), 1310–1316. University of Leeds, 2004. Master Chemical Mechanism. Version 3.1. Retrieved March, 2007, from /http://mcm.leeds.ac.uk/ MCM/S. U.S. Environmental Protection Agency, 1980. Volatile Organic Compound (VOC) Species Data Manual second ed.: EPA450/4-80.-015. U.S. Environmental Protection Agency, 2007. Proposed revisions to ozone national standards. EPA-HQ-OAR-2005-0172. Retrieved July, 2007, from /http://www.epa.gov/groundlevelozone/pdfs/ 20070620_fs.pdfS.