A. J. Heber, J.–Q. Ni, B. L. Haymore, R. K. Duggirala, K. M. Keener

ABSTRACT. Reliable measurements of air quality and emissions at large livestock buildings with inherently large spatial and temporal variations of pollutant concentrations are relatively difficult and expensive. Appropriate methodologies for such measurements are not readily apparent and techniques and strategies vary widely. Several important technical issues need to be addressed by an air pollutant emission measurement plan. This article describes comprehensive field measurements of indoor air quality and air pollutant emissions at eight commercial swine finishing buildings. The objective of the field test was to evaluate the effect of a manure additive on concentration and emission of ammonia, hydrogen sulfide, and odor. Continuous measurements of gases, ventilation rate, building static pressure, inside and outside temperature and humidity, and wind speed and direction were conducted at four naturally–ventilated buildings and four mechanically–ventilated buildings. Air was pumped continuously from inside each building into air–sampling manifolds. One air stream was drawn from beneath the floor to assess pit headspace air concentrations. Another air stream was drawn from ventilation exhaust points to assess inside gas concentrations and building emission rates. Gas analyzers were switched between sampling manifolds on 10– to 15–min sampling intervals. Ammonia was measured with chemiluminescence NOx analyzers after conversion to nitric oxide. Hydrogen sulfide was converted to sulfur dioxide and measured with pulsed–fluorescence, sulfur dioxide analyzers. Odor samples were collected in bags and evaluated using olfactometry. Gas and odor emission rates were determined by multiplying mean gas concentrations in the exhaust air by ventilation airflow rates. Ventilation rates of naturally–ventilated buildings were estimated using sensible heat and carbon dioxide balances. Ventilation rates of mechanically–ventilated buildings were determined by monitoring wall fan operation and directly measuring airflow of some variable–speed pit fans with full–size impeller anemometers. Labor and equipment requirements, pitfalls, problems, and solutions to problems of field studies are discussed. Several recommendations for future studies of this type were developed based on experience gained during this measurement campaign. Keywords. Air quality, Pig house, Environment, Pollution, Instrumentation.


odern livestock production facilities have increased in size resulting in a greater geographical concentration of animals, increased public scrutiny, stricter government regulation, and more nuisance odor complaints. Increased public acceptance of pork production operations may be achieved by reducing odor, gas and particulate matter emissions. Commercial manure additives reportedly mitigate odor production, reduce ammonia (NH3) and hydrogen sulfide (H2S) emis-

Article was submitted for review in October 1999; approved for publication by the Structures & Environment Division of ASAE in July 2001. Presented at the 1998 ASAE Annual Meeting as Paper No. 98–4058. Mention of commercial products is for the benefit of readers and does not infer endorsement or preferential treatment by the authors. Journal Contribution No. 16125 of Purdue University Agricultural Research Programs. The authors are: Albert J. Heber, ASAE Member Engineer, Associate Professor; Ji–Qin Ni, ASAE Member, Technical Director of Agricultural Air Quality Laboratory, Agricultural and Biological Engineering Department, Purdue University, West Lafayette, Indiana; Barry L. Haymore, President, ChemLink International, St. Louis, Missouri; Ravi K. Duggirala, Manager, Enron Energy Services, Houston, Texas; and Kevin M. Keener, ASAE Member Engineer, Assistant Professor, Food Science, North Carolina State University, Raleigh, North Carolina. Corresponding author: Albert J. Heber, Agricultural and Biological Engineering Dept., Purdue University, West Lafayette, IN 47907; phone: 765–494–1214; fax: 765–496–1115; e–mail:

sions, break down solids, and increase availability of manure nutrients. Some laboratory and small–scale field–testing of these manure additives have been reported (Al–Kanani et al., 1992; Patni, 1993; Miner et al., 1995; Andersson, 1995; Bundy et al., 1996; Li et al., 1996; Zhu et al., 1997). In general, these reports have shown mixed results with very few additives proving their effectiveness in achieving the intended benefits. Large–scale field tests were initiated in 1996 to evaluate the effect of a manure additive (Alliance, Monsanto EnviroChem, St. Louis, MO) sprayed into pits of swine finishing houses on the concentration and emission of NH3, H2S, and odor. A total of 62 building–months of gas emission data was collected at four naturally–ventilated (NV) buildings and four mechanically–ventilated (MV) buildings (Heber et al., 1997, 1998b, 2000). Data collected from the MV buildings have been analyzed and reported elsewhere (Heber et al., 1997, 1998a, 2000; Lim et al., 1998; Ni et al., 1999a, 2000a, 2000b). The manure additive resulted in 26% lower (P < 0.05) building emissions of NH3 but did not significantly affect H2S or odor emissions (Heber et al., 1998b). Reliable measurements of air quality data in large livestock buildings that have large spatial and temporal variations of pollutant concentrations are relatively difficult and expensive (Hinz and Linke, 1998). The appropriate methodology needed for making such measurements is not obvious and the techniques used by other researchers vary

Transactions of the ASAE Vol. 44(6): 1765–1778 E 2001 American Society of Agricultural Engineers ISSN 0001–2351 1765

greatly (Phillips et al., 2001). It was recognized that instrument selection, placement, and operation were critical to the success of the field test of the additive. The objectives of this article were to: 1. Describe the approach used to field test a commercial manure additive. 2. Describe buildings, equipment, and methodology used to conduct the test, 3. Report challenges faced and lessons learned from the study.

The industrial research sponsor’s objective for this project was to test the effect of a newly developed product on building air quality and pig performance. Pig performance data required complete growth cycles with sufficient repetition and replication for statistical validity. It was therefore decided to test pig performance at commercial deep–pit finishing buildings for a total of 12 batches of pigs. Tests were to be completed over a period of 12 months by using four test sites simultaneously and three growth cycles per site. Each site consisted of identical treatment and control buildings. Environmental data were monitored continuously during each three to four month growth cycle. This article focuses primarily on air quality, ventilation airflow, and environmental data collection at these sites. High frequency gas measurements at two or three locations in the buildings were chosen to assess air quality. Long–term continuous data collection was deemed very important because of inherently high spatial and temporal variance of gas concentrations. Additionally, it was felt that product effectiveness would be best evaluated by measuring gas concentrations near the source (pit surface) that was treated by the product (sprayed solution). Thus, the top part of the pit headspace was chosen to represent the volume of air nearest the treated gas emission source. Gas concentrations in the exhaust stream were represented by measurements at exhaust ventilation fans of the MV buildings and inside the inverted–V ventilation chimneys of the NV buildings. Gas analyzers were sequenced between control and treated buildings. Using the same analyzers for both buildings was intended to reduce sampling error (table 1). Analyzers were calibrated weekly to biweekly. This calibration frequency was higher than recommended by the instrument manufacturers because of the harsh environment of swine buildings and the desire to assure higher accuracy.
Table 1. Sequence for switching gas analyzers between sampling location groups. Sites Sampling location group Building 1 and 2 Pit headspace Pit headspace Inverted–V chimney Inverted–V chimney Pit fans Pit fans Wall fans Wall fans Treated Control Treated Control Treated Control Treated Control 1 2 3 4

Other building environmental control variables such as static pressure and particulate matter concentration were measured to verify equal environmental control between treated and control buildings. Carbon dioxide (CO2), long considered an indicator of ventilation rate, was also monitored to verify similar ventilation of the control and treated buildings. An extensive effort to monitor airflow was undertaken to allow calculation of continuous gas and odor emission rates and also to account for ventilation, which affects gas concentration by clearance through air exchange. Airflow estimates in the NV buildings were made possible by measuring temperatures and CO2 concentrations for the heat balance and CO2 balance methods, respectively. Additionally, wind speed, and direction along with ventilation openings were monitored for direct calculation of ventilation rate (Bruce, 1978; Zhang et al., 1988). In this study, the sponsor managed the project, made final decisions on all issues, and hired and supervised field research engineers who worked full–time at the sites. A full–time university research associate was responsible for computer data acquisition, data analysis, and report writing. University faculty provided continuing technical advice to the sponsor’s project manager. Installation of measurement equipment at site 1 near Rushville, IN began in June 1996 and collection of gas concentration and temperature data began in August. Installation of instruments at site 2, another pair of buildings at the same farm, began in August and data collection began in December. Four of nine barns at a swine finishing facility near McLean, IL were selected in October 1996 for sites 3 and 4. Equipment installation began in November and was completed by June 1997. Experiments at sites 1 and 2 were discontinued in August 1997. Sites 3 and 4 experiments were discontinued in September 1997.

DESCRIPTION OF SITES Site Selection Requirements The following requirements and criteria were used to select eight commercial swine finishing buildings for sites 1–4: 1. Identical but independent buildings with at least 100 finishing pigs in each building. 2. Buildings separated by less than 30 m. 3. Long–term manure storage with 1.8 to 2.4 m deep pits. 4. Separate manure pits for each building with no communication of liquid between them. 5. All–in all–out production with both buildings filled with pigs within two weeks. 6. Mechanical–ventilation preferred, but not required. 7. Natural ventilation should have either pit exhaust fans or a central chimney outlet. 8. At least 2.0 m2 of floor space available for manure additive application equipment. 9. At least 8.0 m2 of floor space available for air sampling and measurement equipment. 10. Similar management and design, e.g. ventilation, feeding, manure collection, etc.

Sites 3 and 4 1 2

3 4 5 6



Figure 3. Floor plan (top) and side view (bottom) of buildings with instrument placement at sites 1 and 2 (circle = temperature, triangle = air sampling locations in pit headspace, square =air sampling locations in chimney). All dimensions in m.

structed in the northeast corner of the 3.0 × 12.2 m workroom. Air was forced into the instrumentation room through a bag filter to remove dust. Instrumentation had to be protected from manure gases since the pit under the east building extended underneath this workroom. The workroom was typically heated by opening end wall panel doors to each building, allowing cross flow of air through the workroom. However, these panel doors were closed during the test. One curtain controller per building was used to simultaneously raise and lower the south and north curtains to control building temperature. Mechanically–Ventilated Buildings at Sites 3 and 4 Sites 3 and 4 consisted of the southernmost four of nine identical, 1000–head, MV finishing buildings. Buildings 1 through 9 were oriented east–west from north to south, respectively. Site 3 consisted of buildings 6 and 7, and site 4 consisted of buildings 8 and 9 (fig. 4). These buildings (figs. 5 and 6) had the following characteristics: 1. 12.3 m wide and 65.9 m long with capacity for 1000 pigs starting at 25 kg each. 2. Twenty pens (3.3 × 5.8 m each) on either side of a 0.76 m wide central alley. 3. Flat, 2.3 m high ceiling with loose cellulose insulation (0.08 m deep). 4. Totally–slatted concrete floor above a 2.44 m deep ventilated pit (799 m2 surface area). 5. Sidewall curtains (1.5 m high) with one curtain controller per building. 6. Curtain in west end wall for hot–weather tunnel ventilation.

Site 1 Control N Site 2 12.2 Covered walkway Meteorology tower (9 m) Treated 54.9 3.1 Treated Instrument rooms 18.3 (2.0 x 4.6) Control 54.9

Figure 1. Layout of buildings at sites 1 and 2. All dimensions in m.

Vol. 44(6): 1765–1778


Naturally–Ventilated Buildings at Sites 1 and 2 Sites 1 and 2 consisted of four buildings (built in 1995) arranged in an H–layout with an east–west orientation (fig. 1). The finishing site was located in the middle of a field with excellent wind exposure. Sites 1 and 2 were the north and south building pairs, respectively. The northeast and southwest buildings were treated with the manure additive, whereas the northwest and southeast buildings were untreated. Both buildings of each site were filled with 20.5 kg gilts and barrows. These NV buildings (figs. 2 and 3) had the following characteristics: 1. 12.2 m wide and 54.9 m long with capacity for 1000 pigs starting at 25 kg/pig. 2. Totally–slatted concrete floor above a 2.44 m deep pit, no pit ventilation. 3. Twenty 2.75 × 5.80 m pens on either side of a 0.66 m wide central alley. 4. Flat, 2.3 m high ceiling. 5. Sidewall curtains (1.5 m high) with one curtain controller per building. 6. Inverted–V chimney that was 1.2 and 0.2 m wide at the ceiling and ridge, respectively. 7. Ridge closure consisting of a manually–controlled PVC tube. 8. 18.3 m space between buildings. 9. All–in all–out production with routine recording of feed consumed by each pig group. Manure pits under the east buildings extended underneath the workroom. Thus, the total pit surface areas of the east buildings were 683 m2 compared with 669 m2 in the west buildings. Wet/dry feeders were used in these buildings until 1997 when a liquid system was installed for feeding cheese whey. Instrumentation was housed in a clean, 2.0 × 4.6 m, positive–pressure, environmentally–controlled room con–

Figure 2. Cross–section of building and additive spray system at sites 1 and 2 (circle = temperature, triangle = air sampling locations in pit headspace, square = air sampling locations in chimney). All dimensions in m.
20 pens (2.8 x 5.8 m each side) 12.2 T1 5.5 11.0 T2 Air sampling locations 27.4 T12 T7,14 T10,11 T3,16 T9 T6,13 T8

Additive spray T15 (floated 0.08 m below surface) 12.2

11. Accurate producer records of total feed consumption of each group of pigs. 12. Accurate producer records of initial and final pig weights. 13. Availability of site for three consecutive growth cycles or pig groups. 14. Relatively new facilities that represent modern production methods. 15. Easy driving distance from the university campus. The facilities that met all these criteria were extremely rare and criterion 15 was somewhat compromised since transportation to each site required 2.5 h of driving.

Fiberglass insulation (0.14 m thick) Cellulose insulation (0.15 m thick) T6 1.5 T8 1.78 T13 0.51 Air sampling port 3.1

0.20 T1– 5

PVC tube (0.30 m diameter) 2.0 T7 T12 Curtain T14 Concrete (0.15 m thick)

1.0 T16* 1.22 T9

Air sampling port Pressure T10

T11 Psychrometer 0.66



T4 9.1

T5 4.6

27.4 Meteorology tower T2 T6,7 T3 T16 T8,12 T13,14 T15 T4 T5

Attic Room Pit



7. Thirteen, equally–spaced, center–ceiling, ventilation inlets with 2.4 × 0.3 m insulated baffles. 8. 18.3 m space between buildings. 9. Dry feeders with waterers on the partitions. 10. Both buildings of each site were filled with 20.5 kg barrows (building 6) or gilts (buildings 7, 8, and 9). 11. All–in, all–out production with routine recording of feed consumed by each pig group. There were four, 46 cm diameter, variable–speed pit ventilation fans (Model AT18F, Aerotech, Lansing, MI); a 91 cm diameter wall fan (Aerotech Model AT36Z1C); four, 122 cm diameter wall fans (Aerotech Model AT481Z1C), and two propane heaters (3,955 kJ/min rated capacity, Model C225, Automated Production Systems, Assumption, IL) in each building. The pit ventilation fans ran continuously year–round. Two pit fans were located on the north wall, approximately 7.6 m from each end of the building. The pit fans on the south wall were located about 24 m from each end (fig. 5). Uncapped wooden chimneys with dimensions of 3.6 m × 0.8m × 0.8 m were used with the pit fans to minimize wind effects and facilitate use of high–accuracy impeller anemometers (fig. 6). The chimneys were insulated with 19 mm thick, rigid insulation, protected by 11 mm thick OSB waferboard and supported by a 5cm × 5 cm wood frame. The pit fan blew vertically upward and was installed with the top edge
Building 6 Instrument trailer Building 7 18.3 Meteorology tower, 9 m Building 8 Building 9

flush with the top of the chimney. The impeller anemometer was mounted 0.6 m below the top of the chimney. The floor of the chimney was sloped 5:8 toward the building. A 0.54 m × 0.54 m hole was cut in the chimney to match the hole in the stainless steel pit fan annex. The five exhaust fans located on the east end wall of the buildings along with curtains on the west end wall facilitated tunnel ventilation in hot weather. The buildings were typically naturally–ventilated between 18 and 27°C with tunnel ventilation above 27°C and cross–flow ventilation below 18°C. However, the natural ventilation phase was eliminated for this study. A building environmental controller (BEC) (Model ESU.2, Fancom, Panningen, The Netherlands) was used in each building to control operation of the pit and wall fans based on inside temperatures measured with thermistors (Model A5030146, Fancom, Panningen, The Netherlands). Dual Control of Building Ventilation Strategies for dual building control of ventilation at all sites were developed to equalize airflow rates between treated and control buildings. All four sidewall curtains of the building pair were controlled simultaneously at sites 1 and 2 by inputting the average temperature of four sensors, two per building, to the curtain controller. The BECs were also set up at sites 3 and 4 to equalize ventilation rates. The controlling BEC in the treated building operated the eight pit fans among both buildings at the same speed based on the average of eight temperatures, four in each building. Wall fan relay signals from the controlling BEC were extended to the wall fan relays of the remote building. In all cases, switches were installed to allow convenient toggling between single and dual building control modes. Dual building control strategies were abandoned after it became apparent that, although total building airflows could be kept the same, temperature differences between the buildings became too large. This occurred because of the difference in pig inventory and mass, and consequently, total sensible heat production from the pigs. Dual building control did not allow equal airflow per unit pig mass between the buildings. Therefore, the strategy of controlling both buildings based on the combined average temperature is not practical unless sensible heat production from each group of pigs is nearly identical. Another reason why the strategy did not work in the NV buildings is that the two buildings connected at the ends created essentially one very long building. Very large lengthwise variations can occur in a long building when wind is parallel to its long axis. Equipment Cost The total cost of instruments and major equipment needed for this study (table 2) was approximately $259,000. Additional expense was incurred for environmentally– controlled trailers and instrumentation rooms, spare parts and equipment, miscellaneous and consumable supplies, installation, maintenance, and the personal computer used for data analysis. However, the estimated total cost of the major items illustrates the relatively high expense of making accurate emission measurements at livestock buildings. Of course, it is recognized that some lower cost approaches may be equally effective.

Buried conduit

Heated air sampling tubes

Figure 4. Layout of buildings and instrument trailers at sites 3 and 4. All dimensions in m.
7.6 1.7 Manure pit annex (1.92 x 1.68 m) 2.74 20 pens (3.3 x 5.8 m each side) T7 Alley T6 T5 Pressure
8 9’ 9’ 12

N= T8 T13 T9 T4,15,16 T10,12 T11 T3
36’ 18’





Figure 5. Floor plan and side view of the buildings with instrument placement at sites 3 and 4 (circle = temperature, triangle = air sampling locations in pit headspace, square = air sampling locations in pit fan chimneys, inverted triangle = air sampling locations at wall fans). All dimensions in m.
0.81 0.61 Baffled ceiling inlets (13) 3.6 1.5 T8 1.78 0.51 Pit Pit fan annex T13 3.1 Pressure T16 Pit fan Impeller anemometer Insulation (0.19 m thick) Chimney

Figure 6. Additive spray application system at sites 3 and 4 (circle = temperature, triangle = air sampling locations in pit headspace, square = air sampling locations in pit fan chimneys, inverted triangle = air sampling locations at wall fans). All dimensions in m.


Gas heater 0.80 24.7 16.6 23.2 Wall fans Attic Pit fan chimney T6 1.2 T16 Room Pit T7 T5

9 6, 13T14


T4 T8– 12







7 T10 T9 T1– Psychrometer 0.76 Additive spray

T11 T14

T12 Curtain Concrete (0.15 m thick) Slatted floor


T15 (floated 0.08 m below surface) 12.5



Description NH3 converter NO–NO2 –NOx analyzer H2S converter SO2 analyzer CO2 monitor Compressed gas regulators

Table 2. List of major equipment used at the sites. Source, Make, Model and/or Part Number TEI Model 17C TEI Model 42C (precision = 0.5% full scale) TEI Model 340 TEI Model 43C or 45C (precision = 1 ppb) MSA Model 3600 (precision = 2%) Various models (BOC Model BST–15–660–45, AGT Model SGT500–40–330–DK, and CeeKay Model SGT500–15–590–DK) Machine shop constructed, 3.8 cm × 46 cm Machine shop constructed, 3.8 cm × 31 cm Dwyer Model VFA–24–SS, 10 Lpm Cole Parmer P/N U–06621–40 Cole–Parmer P/N L–79200–00 KNF Neuberger Model PU426–N026.3 8.90 Bios International Model AirPro 6000D Biochem Valve Model OST6–S42 Biochem Valve P/N 100T3–S851–M Biochem Valve P/N 100T2–S801–M Various suppliers Various suppliers Ashcroft Model 0.05 SUBD Ashcroft Model IXLDP (accuracy = ±0.25% full scale) Analog Devices No. SNSR–AD592–PRB6CN (accuracy = ±0.5° C at 25° C; range = –25–105° C) Fancom P/N A5030146 (accuracy ±1.0 °C) FanCom Model A5030009.00 King Electronics Antenna Tower, 9 m FanCom EXAVENT FMS 50, Type 1450M (accuracy = 1% full scale) Applied Technologies Model UFM 311/3NTVZ (accuracies = ±0.05m/s speed, ±0.1° direction) Met One Model 014A–L (accuracy = ±0.11 m/s) Met One Model 024A–L (accuracy = ±5°) General Eastern Model EHRHT1–2I–1 (accuracy = ±2% and ±0.3°C, respectively) Aerotech Model AC1949, 10–turn Local electrical supplier Constructed and installed by local carpenters IBM–compatible Pentium 150 Labtech NotebookPro version 9 Measurement Computing Corporation CIO–1602/16 CIO–EXP–32 CIO–DAS–TEMP SSR–RACK24 SSR–IDC–05 SSR–ODC–05 CIO–DIO24 Fancom UPS Systems Models RST31, RS3RP2T EPE Technologies Model 89121–10B

Qty 4 4 4 4 4 24

Cost, US$ 28,000 40,000 12,000 36,000 7,000 12,000

Sampling manifold Mixing manifold Flow rotameter Air filter holders Air pumps Vacuum pump for NH3 Air sampling pumps 6–way solenoid valve 3–way solenoid valve 2–way solenoid NC valve Tubing and fittings Wire Pressure test gages Differential static pressure Temperature sensors Temperature sensors Motorized psychrometers Meteorological towers Impeller anemometers Ultrasonic anemometer Cup anemometer, wind vane Wind direction sensor RH/temperature probes Feedback potentiometers 240 VAC relays Chimneys for pit fans Personal computer Data acquisition software Data acquisition boards Analog input Analog input multiplexer AD592 input Relay interface board 3–32 VDC input relays 0–60 VDC output relays Digital input/output board Serial interface Uninterruptible power supply Battery set Total cost

16 16 16 88 16 4 8 4 18 4 – – 8 8 128 16 8 2 9 1 1 1 2 4 20 16 4 4 4 4 4 6 20 48 2 2 4 2

3,000 2,000 800 6,200 8,000 4,000 4,500 2,400 1,800 1,500 13,000 1,500 1,000 5,200 6,000 1,000 5,000 200 7,500 13,000 400 600 900 300 1,000 5,600 8,000 3,600 3,000 1,000 3,200 1,000 200 500 100 1,000 4,000 2,000 $259,000

Continuous Measurements Gas Concentration The air sampling system at each site was designed to: 1. Utilize one set of gas analyzers to measure gas concentrations in multiple air sample streams, 2. Deliver sample air from each remote location to the gas analyzers in less than 60 s with insignificant losses in accuracy due to surface absorption of gases, air leakage, and residue from previous samples, and

3. Determine daily mean gas concentrations by obtaining at least 16 measurements per location per day (one every 90 min). Air Sampling System. For each sampling location group (SLG), air was drawn equally from multiple air sampling points and mixed into one sample stream for gas measurement. Two air–sampling streams were drawn continuously at about 6.0 Lpm from SLGs inside each building into an environmentally– controlled instrumentation room or trailer. The maximum residence time in the air sampling system for

Vol. 44(6): 1765–1778


any sample was about 45 s. One air stream was a combined flow from six pit headspace locations that were 15 cm beneath the floor (SLG 1). Besides being close to the treated emission source, SLG 1 was considered to be most representative of pig exposure without sampling directly from the pig breathing zone. The other air stream was a combined flow from four to six sampling locations in the exhaust air (SLG 2, SLG 3). A 47 mm diameter, 0.45–µm thick, Teflon membrane filter, placed inside an all–Teflon PFA filter holder (Savilex), was used to remove particulate matter at each sampling location. All filters were replaced monthly. The filter holders at the pit headspace locations were installed in–line about 1.5 m from the end of the tubes for easy accessibility at the top of the pen partitions. This length of tube upstream of the filter was protected from the pigs by a stainless steel pipe. Exhaust air at sites 1 and 2 was drawn from six equally spaced points (SLG 2) along the length of the inverted–V chimney (figs. 2 and 3). At sites 3 and 4, the air sample stream from the pit fan group (SLG 2) was initially the sole exhaust air location. By July, the wall fan group (SLG 3) was sampled by switching between SLG 2 and SLG 3 through sampling manifold M5 with a computer–controlled 3–way valve located in the building (fig. 7). For each air sampling stream, each ported Teflon mixing manifold (3.8 cm o.d., 1.3 cm i.d. × 25.4 cm long, Process Equipment and Controls, St. Louis, MO) (M1 to M3) combined filtered air from four to six locations of an SLG into a main Teflon tube connected to a sampling manifold (M4 or M5, fig. 7). The segment of main tubes exposed to ambient air at Sites 3 and 4 were heated to prevent condensation of water vapor in the air sample stream. The tubes, surrounded by a two–conductor heat trace, were tied to a taut cable suspended between the instrumentation trailer and the buildings (fig. 4). The heat tape was able to keep the tubes at 27°C in 7°C ambient temperature. The mixing manifolds (manufactured in house) were mounted on the ceiling of the test building, and the sampling manifolds (3.8 cm o.d., 1.3 cm i.d. × 45.7 cm long) were mounted vertically adjacent to the gas analyzers. The port for the main air sampling tube (9.5 mm o.d./6.4 mm i.d.) was located 15.2 cm from one end of the sampling manifold. Four sub main ports with 4.8 mm o.d./3.2 mm i.d. Teflon tubes connected at 90° angles from each other were located 7.6 cm from each end of the manifold for a total of eight ports. The exhaust tube was attached to the top of the manifold. A valve at the bottom of the manifold was used to drain condensate as needed. Air samples from SLG 3 were collected 0.5 m directly upstream of the wall fans (figs. 5 and 6). A six–port Teflon manifold (M3) with 2–way solenoid valves on each port (Model OST6–S42, BioChem Valve, Hanover, NJ) controlled fan sampling tubes (6.4 mm o.d./3.2 mm i.d.) from the 122–cm wall fans (the 91–cm fan sampling tube was left open) and mixed sample air from selected wall fans into a main sampling tube (9.6 mm o.d./6.4 mm i.d.) connected to the 3–way valve (fig. 7). When the 3–way valve was switched to SLG 2, about 1.5 Lpm was taken from each pit fan since there were four filtered air–sampling tubes connected in parallel. Airflow calibrators (Gilibrator–2 Bubble Generator with 2–30 Lpm and 0.02–6.0 Lpm flow modules) were used to check airflows and detect leaks in the air sampling system by


Instrument room Wind speed and direction Dry bulb temperature Relative humidity – Differential pressure (2) + Wet bulb temperature (2) Dry bulb temperatures (30) Curtain, ridge positions, Sites 1,2 Wall fan operation, Sites 3,4 ECS serial port, Sites 3,4 P




M1 M4 M2 P S S

NO analyzer NH3– NO conver ter CO2 analyzer SO2 analyzer S H2S – SO2 converter PC Air exhaust


M2 S S S M3 S M2 S

S S M5 M2 S S S P S


Other building* Tubing Wire

Teflon filter

Figure 7. Data acquisition and air sampling configuration for each pair of buildings. Note: the CO2 monitor and SO2 analyzer had internal pumps. P = pump. S = solenoid valve, M = manifold and ECS = environmental control system. *Same scheme for each building.

assuring equal inlet and outlet flows. They were also used to equalize airflows in the sampling tubes leading to each mixing manifold. Each gas analyzer was automatically switched sequen– tially between four sampling manifolds (two per building) to sample air from each SLG. At sites 1 and 2 and initially at sites 3 and 4, gas concentrations of each SLG were measured continuously during a 15–min sampling period before switching to the next SLG. Gas concentrations at each of four SLGs were therefore measured during twenty–four, 60–min sampling cycles per day. There were sixteen, 90–min sampling cycles per day after SLG 3 was added at site 3 (4 June) and site 4 (16 July). The sampling period for each SLG at sites 3 and 4 was reduced to 10 min on 14 August and the daily number of sampling cycles returned to 24. The first three minutes of pre–equilibration gas concentration data during each 15– and 10–min sampling period were discarded. Gas concentration readings averaged over the last 12 and 7 minutes represented the period mean gas concentration. Condensation occurred inside the air sampling tubes in the air–conditioned instrumentation trailers at sites 3 and 4. In certain cases, condensation occurred because the instrumen– tation trailer temperature set–point was less than the dew point temperature in the pig buildings. In other cases, it occurred because cold air from the trailer air conditioner blew on the sampling tubes before mixing well with room air. There were conflicting needs in the instrumentation room. A component in the NH3 analyzer needed to be cool whereas the sampling tubes needed to be warm. With a water bath attachment (available from the manufacturer) to keep the NH3 analyzer cool, the air conditioning could have been turned off or operated with a higher set point so that



condensation in external sampling tubes was prevented. Otherwise, a partition could have been constructed to separate the analyzers and exposed tubes and manifolds. A heater could have blown warm air across the tubing and manifolds behind the partition while the air conditioner blew cold air in front of the partition. There was concern that cold air entering sites 3 and 4 buildings through ceiling inlets may also cause condensation in sampling tubes attached to the ceiling. Vertical deflectors were therefore installed to deflect incoming ventilation air away from sampling tubes. Ammonia. Ammonia was measured by first oxidizing NH3 to nitric oxide (NO) (Phillips et al., 1998) with an NH3 converter (Model 17C, Thermo Environmental Instruments (TEI), Franklin, MA) and then detecting the NO with a chemiluminescence detector in an NO–NO2–NOx analyzer (TEI Model 42C) (table 2). Ammonia in the air sample stream was initially oxidized to NO with a catalytic converter at 875°C. The NO was further oxidized by gas–phase titration with ozone (O3) in the analyzer’s reaction chamber, producing nitrogen dioxide (NO2) in an excited state (eq. 1). At reduced pressure created by a vacuum pump (Model PU426–NO26.3 8.90, KNF Neuberger, Trenton, NJ), some of the excited NO2 molecules emit radiation as they return to a lower energy state. With excess O3, the intensity of this radiation is proportional to the concentration of NO. NO ) O 3 ³ NO 2 ) O 2 ) hv (1)

where O2 is oxygen and hv represents photons, particles of light energy, or radiation energy that is generated by moving electric charges. The emitted radiation was detected by a photomultiplier tube (PMT), which in turn generated an electronic signal that was processed into a gas concentration reading. Sample air was drawn at a flow rate of 0.6 Lpm from the converter into the NH3 analyzer through a particulate filter, a glass capillary, and a solenoid valve. The sample was routed either through a molybdenum converter and a reaction chamber (NOx mode) or through a stainless steel converter and the reaction chamber (Nt mode). Ammonia concentration was calculated based on the difference between these readings. The NH3 analyzer’s measurement range was 5 ppb to 50 ppm. Its precision was 0.5% of full scale and the 0 to 90% response time was 120 s with 10 s averaging. Hydrogen Sulfide. Hydrogen sulfide (H2S) was first converted catalytically at 400°C to sulfur dioxide (SO2) with a H2S converter (TEI Model 340). The converted SO2 was measured with a pulsed fluorescence SO2 Analyzer (TEI Model 45C) according to U.S. EPA Method EQSA–0486–060. This analyzer was based on the principle that SO2 molecules absorb ultraviolet (UV) light, hv1, produced by a high intensity xenon lamp, and become excited at one wavelength (SO2*). The excited SO2 molecules then decay to a lower energy state emitting UV light at a different wavelength, hv2, that is proportional to SO2 concentration (eq. 2). SO 2 ) hv 1 ³ SO* 2 ³ SO 2 ) hv 2 (2) where SO*2 is an excited SO2 molecule, and hv1 and hv2 are emissions of ultraviolet (UV) light. A photomultiplier tube detected UV light emission from decaying SO2 molecules. The SO2 analyzers had a range of

0.05 to 10 ppm, a response time of 60 s (10–s averaging time), and a sample flow rate of 1.0 Lpm. The guaranteed precision was 1% of reading or 1 ppb (whichever is greater). Carbon Dioxide. Carbon dioxide was measured with a photoacoustic infrared gas sensor (Model 3600, Mine Safety Appliances Company, Pittsburgh, PA). The sensor utilized dual frequency infrared absorption and was corrected for water vapor. The range of the CO2 sensor was 0 to 5000 ppm and the guaranteed precision was 2% or ±100 ppm. Sample flow rate was 1.0 Lpm. Gas Calibrations. The gas measurement techniques and sensors described above were chosen for their ability to measure continuously and automatically, and for their reliability, stability, precision, and lack of sensitivity to humidity, especially compared to less expensive electrochemical sensors (Hauser and Fölsch, 1988; Phillips et al., 2001). The choices were based on the goal of obtaining the best possible measurements of air quality. However, gas calibrations are necessary no matter what instruments are used and the accuracy of the instruments are higher if frequently calibrated with high quality standards (Phillips et al., 2001). Gas instruments were calibrated weekly or biweekly with the following certified calibration gases: 1. Pure air, or oxygen (21%) and nitrogen (79%) mixtures 2. Carbon dioxide (3000 ppm) 3. Nitric oxide (24 ppm) 4. Ammonia in air (25 ppm) 5. Sulfur dioxide in air (28 ppb) 6. Hydrogen sulfide in high purity nitrogen (5700 ppb) 7. Oxygen (40%) and nitrogen (60%) mixture The 40/60 mixture of oxygen and nitrogen was used to mix with H2S when calibrating the H2S analyzer. Cylinder pressures were matched to assure proper mixing ratios. A capillary–based flow splitter circuit was used to accurately blend calibration gases to create other concentrations. Airflow Rate Ventilation airflow rates of NV buildings at sites 1 and 2 were estimated with the heat balance method (Hinz and Linke, 1998; Pedersen et al., 1998), which required calculation of heat lost by conduction through the building envelope, heat lost by total ventilation airflow, and sensible heat produced by the pigs. Radiant heat load was assumed to be zero. Building heat loss was estimated using standard heat conduction equations based on a constant exposure factor. The exposure factor was calculated from thermal resistances and building dimensions. Sensible heat produced by the pigs was estimated with predictive equations for animal heat and moisture production (CIGR, 1992) based on pig mass and ambient pig zone temperatures (0.5 m above floor). The equation used to calculate ventilation rate from temperature measurements was: Qv + v where QV v cp Ti To qa * E(Ti * To) c p(T i * To) (3)

= ventilation rate, m3/s = specific volume of air, m3/kg = specific heat of air, J kg–1 °C–1 = inside temperature, °C = outside temperature, °C

Vol. 44(6): 1765–1778


qa = animal sensible heat, W E = exposure factor of building, W/°C Substitute temperatures were used in the case of sensor failures. For example, building temperatures were substituted for pig zone temperatures when both pig zone sensors failed. Heaters in the building generated heat that was not accounted for in equation 3. They also emitted CO2 that would also affect the CO2 balance method of calculating ventilation airflow. Heaters were not monitored. Therefore, the sensible heat and CO2 balance methods were applicable only when heaters were not used in the buildings. Curtain and ridge positions at sites 1 and 2 were monitored with feedback potentiometers (Model AC1949, Aerotech, Inc., Mason, MI). Electronic circuits were constructed to provide 0 to 10 VDC signals that were proportional to curtain openings. However, there was significant drift in these signals due to internal braking systems in the curtain controllers and weekly calibrations were necessary. Pit fan airflow rates at sites 3 and 4 were measured with impeller anemometers (Model EXAVENT FMS 50, Fancom, Panningen, The Netherlands) installed 0.6 m beneath the fans in the chimneys. The rotating speed of each impeller anemometer was recorded every minute by the BEC and the airflow was 1.1 L/s per rpm. The Hall effect was used to generate 4 voltage pulses per revolution of the impeller using an excitation voltage of 12 VDC. Only one pit fan (NE fan) in building 6 of site 3 was equipped with an impeller anemometer and building 7 of site 3 had none. Total airflow rate of the pit fans in building 6 was taken as four times the airflow measured by the NE impeller anemometer since the four pit fans were identical (verified by anemometers in Buildings 8 and 9) and controlled by one BEC. Fan control voltage to the four pit fans in building 7 of site 3 was recorded and airflow was estimated based on an airflow/voltage relationship determined from building 6 data. All four chimneys in each building of site 4 were equipped with impeller anemometers so the total airflow rate was the sum of airflows measured individually. Temperatures were measured to enable the use of the heat balance method at sites 3 and 4 for comparison with the direct method. Operation of each wall fan at sites 3 and 4 was monitored using a solid–state 32–VDC input relay. Five 240–VAC relays were installed near the five fans at the east end of each building and were energized by fan motor voltages. Operation of a fan was sensed by the 240 VAC relay assuming that the motor was not burned out and the belted power train to the impeller was functional. Normally open contacts of the 240–VAC relay were used to switch a 24–VDC signal via multi–stranded cable to 32–VDC input relays in the instrument room. The 24–VDC power supply was mounted on the data acquisition board near the 24–channel solid–state relay rack (Model SSR–24, Measurement Computing Corporation (MCC), Middleboro, MA). The personal computer monitored the state of each relay every second and the mean of the latest 20 readings was recorded every 20 s to disk. Airflow rates of the wall fans were calculated with equations 4 and 5 based on independent fan tests and the measured differential static pressure between indoor and outdoor air. The fans were tested with all accessories in place (Ford et al., 2001).

Q f,122 = – 0.061P + 11.77
where Qf,90 = airflow rate for a 91–cm fan, m3/s Qf,120 = airflow rate for a 122–cm fan, m3/s P = static pressure, Pa.


Gas Generation and Emission Generation and emission rates of each gas were calculated with equations 6 and 7, respectively, for each air sampling cycle. Q G + KQv(C e * C i) Q E + KQvC e (6) (7)

where QG = gas generation rate, mg/s QE = gas emission rate, mg/s QV = ventilation airflow rate, m3/s K = factor to convert ppm to mg/m3 Ce = exhaust gas concentration, ppm Ci = inlet gas concentration, ppm The gas emission rates for the buildings at sites 1 and 2 were based on the mean gas concentration in the ventilation chimney and the total ventilation airflow rate estimated by the balance of sensible heat or CO2. Building gas emission rates at sites 3 and 4 were the sum of gas emissions from the pit fans and the wall fans. Gas emission from the pit fans was determined by multiplying the mean pit fan gas concentration by the total pit fan airflow rate. Gas emission from the wall fans was calculated by multiplying the mean wall fan gas concentration by the total wall fan airflow rate. Concentrations of H2S and NH3 in ventilating air entering the building were assumed negligible. This assumption was confirmed with independent measurements in the upwind ambient atmosphere. However, exhaust air reentry of aerial pollutants at certain other times was probable. Other Continuously Measured Variables Temperature. Air temperatures were measured with semiconductor temperature transducers (two terminal monolithic integrated circuits) encased in 15 cm long stainless steel probes (Model AD592CN, Analog Devices, Norwood, MA). The AD592 transducer was chosen for its range of –25 to +105°C, its accuracy (±0.5°C at 25°C), its immunity to voltage drops and voltage noise over long lines, and its long–term stability. The transducers were located to determine the sensible heat loss of the building through the attic and the walls, and to account for any temperature gradients along the length and width of the building. Temperatures were monitored in the pig occupation zone (2), in the attic (2), in the slurry (1), above the central alley (1), and in the chimney outlet (5) at sites 1 and 2 (figs. 2 and 3). At sites 3 and 4, temperatures were monitored in the pig zone (2), in the attic (1), in the slurry (1), and above the central alley (7) (figs. 5 and 6). The temperature probes in the pig zones were protected by 7.6 cm diameter PVC pipes attached vertically to pen partitions. Sixteen, 12.7 mm diameter ventilation holes were drilled into the pipe to allow air to flow freely past the probe. Plastic–encased AD592 transducers were used to measure pit temperatures. The transducers were placed in a short

Q f,91 = – 0.033P + 6.12




Plexiglas tube filled with two–part epoxy (3M Part EC–1838–LB/A), which was weighted, and embedded in a piece of Styrofoam insulation that could be inserted between concrete slats and floated in the slurry. The temperature transducers were calibrated in a constant temperature bath at installation and approximately every six months thereafter. RTD–type temperature sensors (Model A5030146, FanCom, Panningen, The Netherlands), a component of the BEC system, were placed at quarter points along the building length along either edge of the central alley (two on the north side and two on the south side alternating with the south side first) and 1.5 m above the floor. Two Fancom temperature sensors were added to the two existing sensors per building to bring the total number of sensors to four per building for this experiment. Ventilation fan control was based on the average of the readings from the Fancom sensors. Relative Humidity. Temperature and relative humidity were measured at each research location with a combined probe (Model EHRH/T1–2–I–1, General Eastern, Woburn, MA) with ±0.3°C and ±2.0% accuracies, respectively. A housing was constructed for the RH/temp probe and fitted into a 12–VDC aspirated radiation protection shield (Model 43408, R.M. Young Company, Traverse City, MI), which was placed at a height of 3 m. Radiation errors were claimed to be less than 0.2°C with the shield when exposed to solar radiation of 1100 W/m2. Plastic–encased AD592 transducers were used in a filtered and motorized 24–VDC psychrometer (Model A5030009.00, Fancom, Panningen, The Netherlands) placed in each building (figs. 2 and 6) to continuously monitor wet and dry bulb temperatures (accuracy = ±0.5°C). Air density and relative humidity were calculated during post processing of the data. Static Pressure. A ±30 Pa (accuracy = ±0.25%) differential pressure sensor (Ashcroft Model IXLDP, Dresser Instruments, Stratford, CT) was utilized to measure the static pressure differential between the outside and inside of each building. The static pressure lines consisted of two 6.4 mm i.d. Tygon tubes. One tube extended into the building and was attached to the ceiling about 3 m from the wall nearest the instrumentation room or trailer. Another tube extended outdoors and faced downward about 0.5 m above the ground. A metal cap and male pipe adaptor was attached to the end of each tube. A 1.0–mm hole was drilled into the top of the cap to allow measurement of static pressure while minimizing effects of dynamic pressures (Heber et al., 1991). Zero calibrations were conducted by shunting pressure sensor inputs. Wind Speed and Direction. Wind speed and direction were measured at sites 1 and 2 with a 0.02 to 60 m/s, 3–dimensional, ultrasonic anemometer (Model UFM–311/3NTVZ, Applied Technologies, Boulder, CO) mounted about 10 m above the ground, 9 m away from the building and adjacent to the connecting hallway (fig. 1). Wind speed and direction at sites 3 and 4 were measured with a cup anemometer and wind vane (Models 014A–L and 024A–L, Met One Instruments, Grants Pass, OR) mounted about 10 m above the ground on a telescoping antenna tower (fig. 4). Pulses from the cup anemometer were counted by the data acquisition system using a counter input channel. A

one–turn potentiometer on the wind vane was excited by 5.0 VDC and monitored with an analog input channel. Data Acquisition System Data acquisition systems at each site consisted of a 150–MHz personal computer, a 16–bit, 8–channel input/output board (Model CIO–1602/16, Measurement Computing Corporation (MCC), Middleboro, MA), and a 32–channel temperature measurement board (MCC Model CIO–DAS– TEMP) designed for use with AD592 temperature transducers. A 3.1–KVA uninterruptible power system (Model RST31, UPS Systems, San Diego, CA) and a backup battery set (Model 89121–10B, EPE Technologies, Costa Mesa, CA) was used at each site to condition poor quality AC power and to prevent electric power interruptions from disrupting data collection. The following equipment was mounted on a 90 cm × 45 cm plywood panel adjacent to the computer: 1. 24–VDC, 4.8–A power supply. 2. 5–VDC power supply (sites 3 and 4 only). 3. Two static pressure sensors. 4. One or two 24–channel solid–state relay racks (MCC Model SSR–RACK24) with 32–VDC input relays (MCC Model SSR–IDC–05) and 60–VDC output relays (MCC Model SSR–ODC–05). 5. One digital input/output termination board with 24–channels (MCC Model CIO–DIO24/CTR3) 6. A 32–channel analog input multiplexer (MCC Model CIO–EXP32) 7. Auto/manual DPDT switches (on–off–on) to control operation of solenoid–operated air intake valves for the gas analyzers. Switches 1–4, 5–8 and 9–12 were for the H2S, CO2, and NH3 analyzers, respectively. Data acquisition programs were written using icon–level software (LabTech NotebookPro v. 9, Laboratory Technologies Corporation, Wilmington, MA). The programs for sites 1 and 2 and sites 3 and 4 had only minor differences. The BECs of the control and treated buildings at sites 3 and 4 were looped through a serial interface. Data included pit fan control signals and airflows, inside and outside temperatures, and fan and heater operations (table 3). Another program (F–Central for Windows, Fancom, Panningen, The Netherlands), running simultaneously, communicated with two BECs, and recorded data to disk. Periodic Spot Measurements Particulate Matter Concentration A 1.0 to 6.0 Lpm, constant airflow pump (AirPro Model 6000D, Bios International Corporation, Pompton Plains, NJ), operated with AC electric power, was used to sample particulate matter (PM) onto 37 mm diameter, glass fiber filters (Type A/E Filter 61653, Pall Gelman Laboratory, Ann Arbor, MI) for gravimetric determination of PM concentrations. Filters were purchased pre–loaded into three–piece cassettes (VWR Scientific Products, McGaw Park, IL). Cassette and filter combinations were desiccated for 24–h prior to weighing. A 24–h PM sample was taken weekly in the center of each building near the ceiling. The purpose of the PM measurement was to document expected similarity in PM concentration between buildings.

Vol. 44(6): 1765–1778


Table 3. Information recorded with FanCom F–Central control program at sites 3 and 4. Site 3 Buildings Site 4 Buildings Variable Pit fan control voltage, % max Wall fans (on/off) 1 (91 cm dia.) 2 (122 cm dia.) 3 (122 cm dia.) 4 (122 cm dia.) 5 (122 cm dia.) Impeller anemometers, % max 1 2 3 4 Temperatures 1 2 3 4 5 6 Outside 6 x x x x x x 7 x x x x x x 8 x x x x x x 9 x x x x x x

over a period of 15 min, were analyzed to determine gas analyzer equilibrium time. A normalized concentration ratio Ci was calculated for each measurement by dividing data representing readings 1 to 45 (j = 1 to 45) by the mean of readings 9 through 45. Confidence intervals (95%) of Cj were calculated for j = 1 to 35 (fig. 8). Equilibrium is indicated when the confidence interval approached zero. Confidence intervals of C4, C6, and C10 for CO2, NH3, and H2S were 0.02, 0.04, and 0.10 indicating equilibrium times of 80, 120, and 180 s, respectively. An equilibrium time of 180 s seemed justified based on these data. Data Quantity and Quality Three data sets acquired from each site included: 1. electronic data obtained by the data acquisition systems, 2. animal numbers and weights provided by site managers, and 3. field engineer research notebooks that provided manure depth (weekly), water use (weekly), calibration data, etc. Five different types of electronic files retrieved from sites 1–4 contained the following information: 1. Time, date, gas concentrations, relative humidity, static pressure, and relay control codes (every 20 s). 2. Time, date, and indoor and outdoor temperatures (every 20 s). 3. Outdoor temperature, and wind speed and direction at sites 1 and 2 (every 0.1 s) from the ultrasonic anemometer. 4. Time, outdoor and indoor temperatures, wall fan operation, and chimney fan airflow rates at sites 3 and 4 (every 60 s). These data were acquired through the BEC system (table 3). 5. Time and wall fan operation at sites 3 and 4 (every 20 s). Some files had redundant information, e.g., time, wall fan operation, and temperature. A total of 42 electronic files were retrieved daily from the four sites. The number was higher when daily files were broken into segments due to power failures or shutdowns for maintenance or instrument calibrations. The maximum number of data points among 410 variables retrieved daily at four sites was 1,298,720 for gas, pressure, temperature, relative humidity, and airflow, and 3,456,000 for the ultrasonic anemometer, requiring a total of 83 MB of disk storage. Data quality problems occurred due to disconnections, damage caused by pigs, sensor drift, sensor corrosion, and software bugs (F–Central for Windows). Some measurement errors were detected in about 80% of the temperature data files. Pit temperature sensors were especially unreliable.
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 5 10 15 20 25


x x x x x x x x x x x x x x

x x x x x x x x x x

x x x x


Odor Concentration Approximately weekly, odor samples were collected in 80–L Tedlar bags and sent overnight to Iowa State University for analysis by dynamic dilution olfactometry. A bag was filled by placing it into an evacuation chamber and pumping out the air surrounding the bag using a 5–Lpm air pump. The resulting negative pressure inside the chamber caused the bag to inflate with sample air through a 3.1 mm i.d. Teflon tube. The odor sampling scheme at sites 3 and 4 was modified several times as follows: 1. 7 April to 18 May (n = 38): Two odor samples were collected from the sampling manifold for SLG 2. The bag–sampling pump needed to be capable of overcoming relatively large negative pressure in the manifold. 2. 19 May to 22 June (n = 20): Two samples were collected from a location 3 m from the east end of the building and 15 cm above the floor. 3. 23–26 June (n = 4): Two samples were collected from a location 30 cm in front of the inlet to the 91–cm fan. 4. 26 June to 6 August (n = 36): One sample was taken from inside the northeast pit fan chimney 1.4 m above the floor of the chimney. Another sample was taken from a location 30 cm in front of the inlet to the 91–cm fan. 5. 7–18 August (n = 4): Two samples were taken at the 91–cm fan and two samples from the northeast pit fan chimney as described in (4) above. 6. 19–26 August (n = 7). Pit fan samples were collected through Teflon tubing extending from the pit fan chimney inlet to the central alley inside the building. DATA ANALYSIS AND QUALITY Equilibrium Time Ninety–three gas measurements taken at site 4 on 2 August, each consisting of forty–five, 20–s readings taken

Confidence Interval

Ammonia Hydrogen sulfide Carbon dioxide

Number of 20–s Readings Figure 8. Confidence interval of normalized gas levels for 93 measurements taken on 2 August at site 4.



Systematic outliers in gas concentration data, created whenever instruments were calibrated, were deleted manually during data processing.

GAS CONCENTRATIONS AT SITES 1 AND 2 The NV buildings at sites 1 and 2 had two ventilation outlets when wind blew perpendicular to the building: 1. the inverted–V chimney, and 2. the downwind sidewall curtain opening. The chimney was ultimately chosen as the best location to measure exhaust gas concentration and inside air quality because it had the following characteristics and advantages: 1. The width was 1.22 m, or only 10% of the building width. 2. It was located near the human breathing zone and only 1.5 m above the pigs. 3. It was located the greatest possible distance from the sidewall inlets, which could have potentially diluted the sample with fresh air and given unrepresentative and unrealistically low readings. 4. It was located the greatest possible distance from noxious gas sources (fresh dung, manure pit, animal respiration, and floor litter) that would have given unrealistically high readings if sampling too close to them. 5. It “sampled” the average condition of inside air during wind–induced cross flow. 6. It was located where air was most well mixed the greatest percentage of time. 7. It was the sole exhaust of air a large percentage of time during cold weather and calm conditions. 8. If managed properly, it would always exhaust air rather than admit air. The quality of air flowing upward through the chimney is theoretically the same along the length of the building. However, spatial variability of air quality near the floor is inherently very large. It would require numerous individual measurements at several locations to assess spatial variance. To reduce experimental costs, air from six points along the length of the chimney were merged into one monitored air stream that represented chimney exhaust air. AIRFLOW ESTIMATION AT SITES 1 AND 2 The use of an artificial tracer gas such as SF6 is much preferred over the heat balance or CO2 method (Phillips et al., 2001). The choice of the latter was based on experimental costs and, in hindsight, perhaps was ill–advised. The heat balance in a large livestock building is not steady–state and there are significant sinks and sources in the mass of concrete and other building materials. The lack of accounting of solar heat gain during the day and radiation heat loss at night, and lack of validated sensible heat production equations for pigs likely introduced significant errors into airflow estimates based on heat balance. The CO2 production is also highly dependent on diet and time of day (Pedersen et al., 1998), and CO2 produced by stored manure in the pit can be significant (Ni et al., 1999b). FIELD STUDIES Field studies are often desired by product manufacturers to test the efficacy of their technology in “real–world”

conditions. In the case of livestock confined feeding, it is often perceived that producers have greater confidence in results of field studies as compared with laboratory studies because technology is tested in the “real–world.” However, field studies are relative expensive, time–consuming, and fraught with difficulty in achieving proper experimental controls. Field studies usually involve voluntary collaboration by commercial producers, and regular communication should be established with them throughout the study. This communication should include regular progress reports, sharing of data, documentation of site visits, assurance of disease control measures, assurances of confidentiality, and provision of reports and publications resulting from the study. Regular meetings facilitate good working relationships and knowledge about any new and unforeseen changes in production plans. For example, the producer at sites 1 and 2 unexpectedly began feeding cheese whey with a liquid feeding system midway through the study. PROJECT MANAGEMENT Project directors must assure proper on–site experimental management. Two full–time research engineers were employed in this study to conduct experiments and operate prototype chemical application systems. They were trained in standard operating procedures developed by project management. One engineer at each location operated two sites with limited assistance. They would have benefited from daily part–time help to enable timely completion of duties. A significant amount of time (10–15%) should be budgeted for maintenance of equipment after unforeseen breakdowns. A full–time university research associate analyzed data and spent a significant amount of time helping the field engineers with on–site installation, calibration and maintenance. Data processing was also very time–consuming. The time required (12 months) to develop the data processing and analysis tools to facilitate ongoing analysis of data with weekly reporting was a source of frustration for the sponsor. PROBLEMS AND SOLUTIONS The following problems occurred during this field experiment. 1. Condensation of water vapor in air sampling tubes caused problems with the NH3 analyzer. Maintaining temperature of the air sampling tubes at least 3°C above the temperature of sampled air prevented further condensation. 2. Sensors failed and drifted. A detailed quality assurance project plan and more stringent quality control procedures would have minimized this problem. For example, the distilled–water reservoir in the psychrometers frequently ran dry causing errors in wet bulb temperatures. 3. The chemiluminescence analyzer’s reaction chamber acquired deposits of ammonium nitrate causing loss of calibration. Regular cleaning was necessary (Phillips et al., 1998). 4. The computer hard drive failed. Daily backup of data minimized the consequence of failure. A spare hard drive in anticipation of this problem would have minimized interruption of data collection.

Vol. 44(6): 1765–1778


5. Dust and moisture accumulations in air sampling tubes and filters. This problem was solved by replacing filters based on fixed schedules rather than observations. 6. Pigs damaged several sensors and tubes. Solutions include keeping sensors and tubes well out of reach of the pigs, and “over designing” protection devices. 7. Harsh environments caused sensor corrosion resulting in electrical disconnections and bad data. Providing robust protection of exposed electrical wires and connections is critical. Periodic sensor replacement may also be necessary. 8. Air leaks developed in seams between pit fan chimneys and metal annexes. Better post foundations for the chimneys were needed to prevent excessive settling. 9. Pit fans failed frequently due to faulty circuit breakers that were later replaced by the manufacturer. Detection of pit fan failures could have been monitored using current sensing relays on the electrical supply wires of each fan motor. 10. Lightning strikes caused electronic damage. Data collection/storage systems are more vulnerable to lightning and stray voltage in rural areas than on university campuses. Data integrity can be assured by proper backup procedures and uninterruptible power supplies. 11. Problems with BEC software (F–Central) resulted in bad data files. The BEC manufacturer could have solved this problem by troubleshooting their software. Farm personnel mismanaged the side wall curtains of the NV buildings during cold and mild weather. They maintained the south curtain opening 10 cm wider than the north curtain opening and tended to shut both sidewall curtains completely during cold weather. Air would flow in and out of the inverted–V chimney during the winter when both sidewall curtains were completely closed. This phenomenon interrupted the consistent upward flow of exhaust air out of the chimney that was required by the experiment. To solve this problem, the top of the curtain was straightened so that it could be closed to a relatively consistent 2–cm opening along its full length. University staff attempted to convince the producer to allow both curtains to remain at least 2 cm open even in the coldest weather.

Comparison between several techniques of determining ventilation airflows in this project is the subject of future data analysis. These techniques include fan operation monitoring, impeller anemometers, CO2 balance, sensible heat balance, and monitoring of wind velocity and curtain openings. Air sampling to assess animal exposure should be located in the animal’s breathing zone. In this study, air sampling was below the slatted floor, but the floor served as a “leaky barrier” between the pit headspace and the animal zone. This barrier tended to separate the two air volumes and prevent free exchange of air between them. The obvious problem with sampling air from the animal zone is equipment damage by pigs. However, it is possible to enclose and protect the sampling tube and filter with a heavy wire mesh cage. It is recommended that the filter be located 50 cm above the floor inside a 30 × 30 × 30 cm wire cage with very high porosity, e.g. 75%. This would allow the filter to be at the very end of

the sampling tube instead of 1.5 m from the end of the tube. This approach would have prevented gas adsorption and desorption from deposited dust inside the length of tube upstream of the filter. Sequential air sampling between several locations involves decisions about measurement frequency and duration. Low frequency sampling may not obtain accurate means of gas concentrations in a multi–tube air sampling system with three or more sampling tubes. For example, if there are four tubes and sampling frequency is one period/day, then each tube will be sampled continuously for 6 h/d and diurnal 24–h patterns (low and high concentrations during day and night, respectively) (Ni et al., 2000b) will be poorly represented. Higher frequency sampling adequately represents the original signal pattern if sampling frequency is at least twice as high as that of the original signal. Theoretically, it can provide greater accuracy of gas concentration means and its real–time behavior (Guan and Xia, 1992). However, this theory does not account fully for data discarded while gas analyzers attain equilibrium. The decision to change the sampling period from 15 to 10 min decreased the total duration of usable measurement data at a particular location from 192 to 168 min. However, higher frequency resulted in more information about dynamic behavior of gases by increasing the number of daily measurements from 16 to 24. In this study, it was concluded that mean concentrations at each location were more important to testing product efficacy than dynamic behavior. The inclusion of incoming ventilation air as one of the air sampling locations is recommended to account for exhaust air reentry and to assure accurate measurements of gas production rates in the buildings (Hinz and Linke, 1998). Regular observations of dunging patterns on the floor would also be beneficial since the percentage of floor covered with urine significantly affects NH3 emissions (Ni et al., 1995). Pig weight and feed consumption data provided by producers was not accurate enough to detect relatively small differences in pig performance. More intervention was needed in this study to obtain accurate pig weights. One of the difficulties was that buildings were sometimes emptied over two to four weeks. A continuous feed intake and pig weight monitoring system installed in two randomly selected pens per building would have provided more reliable pig performance data. Heater operations should be monitored continuously especially if heat balance calculations of airflow are desired. Selection of reliable and low– maintenance equipment and sensors are critical for high quality data collection (Ni et al., 1995). Remote computer access to the data acquisition computers at each site is also recommended to improve management of data collection, to minimize biosecurity risk by reducing the number of farm visits, and to facilitate more frequent data retrieval. Careful planning of comprehensive field tests is critical to success, but may be too time–consuming in the minds of enthusiastic sponsors who desire to have quick experimental results. Industrial sponsors of field research should understand false economies of rushing experimental set up and preliminary data collection. Instrumentation should be completely tested, evaluated, and fine–tuned in the convenience of the laboratory before taking it to remote field sites. Although it provided hope to an anxious sponsor, the configuration of instrumentation at the sites was very inefficient. It would have ultimately saved time to attain



basic functioning of the instrumentation, air sampling, and data acquisition systems prior to field installation. In retrospect, more time should also have been budgeted for onsite installation, testing, and optimizing of equipment, and development of data analysis software. Successful field studies will allow ongoing modification of experimental protocols based on feedback from initial or ongoing results.

speed and direction. Particulate matter samples were collected weekly and analyzed gravimetrically. Odor samples were collected in bags and evaluated using olfactometry. Labor and equipment requirements, pitfalls, and solution to problems of field studies of air quality were discussed. Several recommendations for future studies of this type were developed based on experience gained during this measurement campaign. ACKNOWLEDGEMENTS Financial support from Monsanto EnviroChem and the Purdue Agricultural Research Programs is acknowledged. The cooperation of PremaLean Pork and Heartland Pork is also acknowledged. The assistance of Dr. Alan Sutton and Dr. Dwaine Bundy, Mark Spence, Claude Diehl, Scott Brand, Garry Williams, Dan Kelly, and Slava Adamchuk was appreciated.

A field test was conducted to evaluate the effect of a manure additive used in large swine finishing houses on concentration and emission of NH3, H2S, and odor. Four test sites or pairs of identical buildings were chosen based on strict selection criteria. Sites 1 and 2 were located at a finishing site where four naturally–ventilated buildings were arranged in an H–layout. The buildings had inverted–V ventilation chimneys, sidewall curtains, flat ceilings, totally– slatted floors, and unventilated deep pits. Sites 3 and 4 consisted of four mechanically–ventilated, deep–pit, finishing buildings. Four continuously–operating variable–speed pit fans along with high–accuracy impeller anemometers were installed in vertical chimneys. Five additional wall fans provided mild and hot weather ventilation. Thirteen, static–pressure–controlled, center– ceiling inlets were used during cold and mild weather. Tunnel ventilation was utilized in hot weather. Gas measurements utilized a mixing manifold to mix filtered air from two to three sets of four to six individual air–sampling tubes in each building. Mixed air was pumped continuously into sampling manifolds. One air stream was taken from six points under the floor to assess pit headspace concentrations. The other air stream was used to obtain gas concentrations in exhaust air to assess indoor air quality and gas emission rates. The gas emission rate at any time was the product of mean exhaust gas concentration and total airflow rate. A computer at each site controlled sequential switching of NH3, H2S, and CO2 analyzers between four sampling manifolds (two per building) on 10– or 15–min sampling intervals. Data collected during the last 7 or 12 minutes were averaged to determine gas concentration. Ammonia was measured with chemiluminescence NO– NO2–NOx analyzers after conversion to nitric oxide. Hydrogen sulfide was converted to SO2 and measured with pulsed–fluorescence, sulfur dioxide analyzers. Carbon dioxide was measured with photoacoustic infrared sensors. Gas instruments were calibrated weekly or biweekly. Ventilation rates of the NV buildings were determined with the heat balance method that involved calculating heat lost by conduction, ventilation heat loss, and pig sensible heat production. Heat production equations based on pig weight and ambient temperatures measured in the pig zone were employed. A CO2 balance for estimating airflow (Pedersen et al., 1998) was made possible by monitoring CO2. Pit fan airflow rates were measured at sites 3 and 4 with impeller anemometers. On–off operation of each wall fan was monitored using a relay activated by fan motor supply voltage. Data acquisition systems continuously recorded air temperatures, building static pressure, outside relative humidity and temperature, inside wet bulb temperature, and wind

Al–Kanani, T., E. Akochi, A. F. MacKenzie, I. Alli, and S. Barrington. 1992. Odor control in liquid hog manure by added amendments and aeration. J. Environ. Qual. 21: 704–708. Andersson, M. 1995. Performance of additives in reducing ammonia volatilization from cow slurry. In Agricultural and Food Processing Wastes: 7th International Symposium, 583–589. St. Joseph, Mich.: ASAE. Bruce, J. M. 1978. Natural convection through openings and its application to cattle building ventilation. J. Agric. Eng. Res. 23:151–167. Bundy, D. S., J. Zhu, S. J. Hoff, and Q. Liu. 1996. Development procedures and experiences in evaluating manure additives for odor control. In Air Pollution from Agricultural Operations: 1st International Conference, 271–273. Ames, Iowa: Midwest Plan Service, ISU. CIGR. 1992. Report of Working Group on Climatization of Animal Houses, 2nd ed. Ghent, Belgium: Center for Climatization of Animal Houses, Advisory Services, Faculty of Agricultural Sciences, State University of Ghent. Ford, S. E., G. L. Riskowski, and L. L. Christianson. 2001. Agricultural Ventilation Fans: Performance and Efficiencies. Bioenvironmental and Structural Systems Laboratory, University of Illinois, Urbana–Champaign, Ill. Guan, Z. Z., and G. L. Xia. 1992. Signals and Linear Systems, Third Edition. Beijing: Higher Education Press. (In Chinese). Hauser, R. H., and D. W. Fölsch. 1988. Methods for measuring atmospheric dust, micro–organisms and ammonia in poultry houses: review and approved practices. J. Vet. Med. B35: 579–593. Heber, A. J., S. E. Cole, and J. P. Murphy. 1991. Facility for testing exhaust ventilation fans. Appl. Eng. in Agric. 7(5): 599–605. Heber, A. J., R. K. Duggirala, J. Ni, M. L. Spence, B. L. Haymore, V. I. Adamchuk, D. S. Bundy, A. L. Sutton, D. T. Kelly, and K. M. Keener. 1997. Manure treatment to reduce gas emissions from large swine houses. In: Ammonia and Odour Control from Animal Production Facilities: International Symposium, eds. J.A.M. Voermans and G. J. Monteny, 449–458. Vinkeloord, The Netherlands, 6–10 October. Heber, A. J., D. S. Bundy, T. T. Lim, J. Q. Ni, B. L. Haymore, C. A. Diehl, and R. K. Duggirala. 1998a. Odor emission rates from swine finishing buildings. In Proc. Animal Production Systems and the Environment, 305–310. Des Moines, Iowa, 19–22 July. Heber, A. J., J. Q. Ni, T. T. Lim, and A. L. Sutton. 1998b. Field tests of Alliance. West Lafayette, Ind.: Purdue University. Heber, A. J., J. Q. Ni, T. T. Lim, C. A. Diehl, A. L. Sutton, R. K. Duggirala, B. L. Haymore, D. T. Kelly, and V. A. Adamchuk.

Vol. 44(6): 1765–1778


2000. Effect of a manure additive on ammonia emission from swine finishing buildings. Trans. ASAE 43(6):1895–1902. Hinz, T., and S. Linke. 1998. A comprehensive experimental study of aerial pollutants in and emissions from livestock buildings. Part 1: Methods. J. Agric. Eng. Res. 70:111–118. Li, X. W., D. S. Bundy, J. Zhu, and N. Rashid. 1996. The effects of manure additives and crust on offensive odor control. ASAE Paper No. 96–4095. St. Joseph, Mich.: ASAE. Lim, T. T., A. J. Heber, J. Q. Ni, R. Duggirala, B. L. Haymore, and C. A. Diehl. 1998. Production of carbon dioxide in a deep–pit swine finishing house. ASAE Paper 98–4049. St. Joseph, Mich.: ASAE. Miner, R., D. Godwin, P. Brooks, W. Rulkens, and C. Kielich. 1995. A protocol to evaluate the effectiveness of odor control additives. In Agricultural and Food Processing Wastes: 7th International Symposium, 271–283. St. Joseph, Mich.: ASAE. Ni, J. Q., D. Berckmans, J. Coenegrachts, and C. Vinckier. 1995. Field measurement of ammonia emission from pig house to collect data for model validation. ASAE Paper 95–4421. St. Joseph, Mich.: ASAE. Ni, J. Q., A. J. Heber, C. A. Diehl, T. T. Lim, R. K. Duggirala, and B. L. Haymore. 1999a. Characteristics of hydrogen sulfide concentrations in two mechanically–ventilated swine buildings. ASAE Paper 99–4155. St. Joseph, Mich.: ASAE. Ni, J. Q., C. Vinckier, J. Hendriks, and J. Coenegrachts. 1999b. Production of carbon dioxide in a fattening pig house under field condition. II. Release from manure. Atmos. Environ. 33(22): 3697–3703. Ni, J. Q., A. J. Heber, C. A. Diehl, and T. T. Lim. 2000a. Ammonia, hydrogen sulfide and carbon dioxide release from pig manure in under–floor deep pits. J. Agric. Eng. Res. 77(1): 53–66.

Ni, J. Q., A. J. Heber, T. T. Lim, C. A. Diehl, R. K. Duggirala, B. L. Haymore, and A. L. Sutton. 2000b. Ammonia emissions from a large mechanically–ventilated swine building during warm weather. J. Environ. Qual. 29(3): 752–758. Patni, N. K., and P. Y. Jui. 1993. Effectiveness of manure additives. ASAE Paper 93–4021. St. Joseph, Mich.: ASAE. Pedersen, S., H. Takai, J. O. Johnsen, J. H. M. Metz, P. W. G. Groot Koerkamp, G. H. Uenk, V. R. Phillips, M. R. Holden, R. W. Sneath, J. L. Short, R. P. White, J. Hartung, J. Seedorf, M. Schröder, K. H. Linkert, and C. M. Wathes. 1998. A comparison of three balance methods for calculating ventilation rates in livestock buildings. J. Agric. Eng. Res. 70: 25–37. Phillips, V. R., M. R. Holden, R. W. Sneath, J. L. Short, R. P. White, J. Hartung, J. Seedorf, M. Schröder, K. H. Linkert, S. Pedersen, H. Takai, J. O. Johnsen, P. W. G. Groot Koerkamp, G. H. Uenk, R. Scholtens, J. H. M. Metz, and C. M. Wathes. 1998. The development of robust methods for measuring concentrations and emission rates of gaseous and particulate air pollutants in livestock buildings. J. Agric. Eng. Res. 70: 11–24. Phillips, V. R., D. S. Lee, R. Scholtens, J. A. Garland, and R. W. Sneath. 2001. A review of methods for measuring emission rates of ammonia from livestock buildings and slurry or manure stores, Part 2: monitoring flux rates, concentrations and airflow rates. J. Agric. Eng. Res. 78(1): 1–14. Zhang, J., K. A. Janni, and L. D. Jacobson. 1988. Modelling for natural ventilation induced by thermal buoyancy and wind. ASAE Paper No. 88–4005. St. Joseph, Mich.: ASAE. Zhu, J., D. S. Bundy, X. Li, and N. Rashid. 1997. Controlling odor and volatile substances in liquid hog manure by amendment. J. Environ. Qual. 26: 740–743.



To top