Particulate Matter (PM₁₀) Concentrations and Emissions at a Commercial Laying Hen House with High-Quality and Long-Term Measurement

Abstract

Particulate matter (PM) is a significant air pollutant in modern egg production. However, high-quality PM data from commercial egg farms are still very limited. A 6-month study, covering both cold and hot seasons, measured PM₁₀ concentrations and emissions in a 140,000-hen commercial laying hen house in the Midwest USA. An advanced measurement system was implemented for continuous and real-time monitoring, collecting data from 67 online instruments and sensors. The study generated 4318 h of valid PM₁₀ data, with 97.8% data completeness. The average daily mean (ADM) PM₁₀ concentration in the house exhaust air, standardized to 20 °C and 1 atm, was 236 ± 162 (ADM ± standard deviation) µg m⁻³. The ADM net PM₁₀ emission was 18.9 ± 2.2 mg d⁻¹ hen⁻¹. Increasing outdoor temperatures were correlated with decreased indoor PM₁₀ concentrations but increased overall emissions. Comparison with the ADM emission of 12.4 ± 13.3 mg d⁻¹ hen⁻¹ from the same house during a previous six-month study in 2004–2005 revealed that artificial hen molting in this study increased PM₁₀ concentrations and emissions. Extrapolating the ADM PM₁₀ emission from the house, the ADM PM₁₀ emission from the entire egg farm was estimated at 35.6 ± 31.1 kg d⁻¹ (or 35.6 ± 4.5 kg d⁻¹ with a 95% confidence interval). This study provides valuable insights into air quality in animal agriculture and contributes high-quality and real-world data for use in data-driven approaches such as artificial intelligence, machine learning, data mining, and big data analytics.

1. Introduction

The rapid growth of the global population in recent decades has significantly increased the demand for food. As a result, global egg production tripled between 1980 and 2020, rising from 26.2 million to 86.7 million metric tons [1]. However, egg production facilities are emission sources of multiple aerial pollutants, including particulate matter (PM), which is among the most concerning air pollutants in modern animal production and is one of the seven criteria pollutants identified in the U.S. EPA Clean Air Act [2].

Particulate matter contains microscopic particles that can be inhaled, potentially causing serious health problems [3]. Many lines of evidence pointed to the fact that exposure to PM was linked to adverse respiratory and cardiovascular health effects [4]. However, the collective evidence could not yet isolate factors or sources that would be closely and unequivocally more strongly related to specific health outcomes [5]. Large quantities of PM emissions can have a negative impact on the environment and ecosystems [6,7].

In poultry houses, most on-farm airborne PM originated from feathers, ranging from 4% to 43% in fine PM and from 6% to 35% in coarse PM, and manure, ranging from 9% to 85% in fine PM and from 30% to 94% in coarse PM [8]. In another study, manure was found to be the main component of broiler chicken house PM [9]. Differences in environmental exposures to PM may impact respiratory outcomes of poultry farm workers [10]. High indoor PM concentrations can also affect poultry health [11].

Data-driven techniques such as machine learning, data mining, big data analytics, and artificial intelligence are becoming increasingly important in studies on environmental pollution from animal farming, e.g., [12,13]. These methods depend heavily on reliable and extensive real-world datasets for model training, improving accuracy, and generating insights. However, there is a severe data gap in air pollution from animal agriculture [14]. Compared with numerous large-scale egg production operations worldwide, the scope of field investigations to determine PM concentrations and emissions is still highly limited. The number of real-world laying hen houses that have been monitored for high-quality data is minimal. This limitation poses a great challenge in an era where scientific research increasingly relies on data-driven approaches.

The qualities and characteristics of PM concentrations and emission rates depend on the number and age of hens, hen density, feed characteristics, ventilation control, weather conditions, lighting, and production practices. Additionally, building design and manure handling also affect pollutant emissions from poultry houses [15].

Manure belt laying hen houses are a relatively newer design, and many were retrofitted from high-rise houses over the past few decades [16]. These houses generally have better indoor air quality than high-rise houses and have been reported with lower PM₁₀ (10 µm particles and smaller) emission rates than high-rise houses in the U.S. [17]. It was demonstrated that more frequent manure removal from manure belt houses can greatly reduce aerial pollutant concentrations and emissions compared with the high-rise houses [15,17,18].

Emissions of aerial pollutants, including PM, from commercial egg production facilities have demonstrated considerable diurnal and seasonal variations. Therefore, long-term (e.g., ≥6 months to cover seasonal variations) and continuous (or high-frequency to cover diurnal variations) monitoring is needed to obtain reliable emission factors and in-depth knowledge about the emission characteristics in field conditions.

The capability for long-term high-frequency emission measurements in livestock housing was facilitated by the development of a common protocol in the U.S. [19]. This protocol was further developed and applied to the National Air Emission Monitoring Study (NAEMS) in the U.S. In the NAEMS, pollutant emissions from 38 commercial livestock and poultry buildings were continuously monitored for two years between 2007 and 2010 using state-of-the-art methodologies and technologies [20]. Eight of the thirty-eight buildings were large laying hen houses, of which six were high-rise houses at three commercial farms in California, Indiana, and North Carolina (e.g., [20]). Two other laying hen houses in the NAEMS were manure belt houses [17].

Investigations on PM in commercial laying hen houses, but with different house structures, measurement methodologies, or monitoring durations, were also reported in Australia [21], Canada [22], China [23], Germany [24], Italy [25,26], Spain [27,28], The Netherlands [29], and USA [30,31,32,33]. However, none of these previously reported cases repeated PM monitoring at the same laying hen house with consistent egg production practices across multiple years to reveal temporal variations in emissions under changing weather and environmental conditions.

This study aims to quantify and characterize PM₁₀ concentrations and emissions through a real-world investigation of a manure belt laying hen house in 2021. It serves as a follow-up to a previous study conducted in the same house during 2004–2005 using the same monitoring protocol. This provided a unique opportunity to compare PM emission baselines 16 years apart. The research was based on high-quality, long-term, and continuous monitoring, providing valuable data to address critical knowledge gaps in understanding PM₁₀ dynamics in commercial poultry operations.

2. Materials and Methods

2.1. The Monitoring Site

2.1.1. Egg Farm and Laying Hen House

The research was conducted on a commercial egg farm located in the Midwest USA. The farm had 14 laying hen houses, which had been converted from a high-rise design to a manure belt design, oriented north to south. Manure produced in the cages dropped on the beneath-cage plastic belts. These belts operated in two groups (cage rows 1–3 and cage rows 4–6) and ran 20% total length each weekday along the house. The manure on the belts was conveyed out of the house and stored in two designated manure storage barns on the farm.

The house selected for PM₁₀ monitoring was identical to the other 13 laying hen houses on the farm. It measured 15.24 m width × 152.4 m length × 7.62 m ridge height 4.9 m sidewall height. It was a two-story structure with six rows and seven tiers of cages (Figure 1). The two stories were separated with grated walkways. Concentrations and emission rates of PM₁₀ were studied for six months, from 16 March to 15 September 2021, spanning cold and hot seasons.

2.1.2. House Ventilation

The house had 46 single-speed belt-drive wall ventilation fans, half in the west sidewall (fans 1–23) and another half in the east sidewall (fans 24–46). The fans were 1.12 m diameter fans (RayDot, Cokato, MN, USA) with 1.32 × 1.32 m fan boxes. The distance between the fans was 6.48 m. The exhausts of the fans did not have discharge cones but had exhaust hoods attached to the sidewall over each fan. The hood measured 3.05 m in height and 1.60 m in width and had a bottom depth of 1.37 m (top depth of 0.00 m), providing an opening area of 2.192 m² at the bottom of the air exhaust hood (Figure 1).

Outdoor fresh air was drawn into the house from the eave openings through the attic and ceiling air inlets. Air inlets were designed on the ceiling in two rows and three sections along the house. They were equipped with 0.42 m width adjustable baffles, which were controlled by wire and motor with a maximum drop of 0.22 m. More details about the ventilation were presented by [34].

2.1.3. Laying Hens

The house had 140,339 W36 Hyline laying hens of 36 weeks old on the first monitoring day (16 March 2021). By the end of the monitoring period on 15 September 2021, the total inventory was 136,041 hens, a decrease of 4298 hens or 3.06% of the initial number of hens. The weekly mean mortality rate during the 26.1-week period was approximately 0.12%, a little higher than the rate reviewed by [35], who summarized that mortality in a healthy, well-managed flock housed in conventional cages is generally less than 0.1% per week. The egg farm had 1.689 million hens in 11 houses on 16 March. By the end of the PM₁₀ monitoring (15 September), 1.973 million hens occupied 13 of the 14 laying hen houses.

The average weights of hens in the monitoring house and the egg farm were 1.50 ± 0.05 kg and 1.54 ± 0.05 kg (mean ± standard deviation, unless otherwise explained), respectively, during the 6-month PM₁₀ monitoring period. The recorded hen weights were updated approximately weekly. Artificial molting was conducted from July to September. During molting, the average weights of hens first dropped, then recovered gradually, and finally exceeded the weights prior to molting by September.

2.2. Monitoring Method

2.2.1. Mobile Lab

A 2.44 m width × 7.32 m length × 2.74 height mobile lab was used to conduct the on-farm PM₁₀ emission monitoring. The same mobile lab technique had been used in the NAEMS project. The mobile lab was equipped with PM monitors, various sensors, and a computer data acquisition and control system. All the instruments and sensors were cable-connected to the data acquisition system, except for a weather station, which was wirelessly connected.

The mobile lab was parked 10 m from the west sidewall of the house and was secured to the ground with four tie-down anchors. It was connected to the house with the raceway (Figure 1). One end of the raceway went through the wall on the south side of fan 5.

2.2.2. Monitoring Instrument and Sensors

Sixty-seven different measurement instruments and sensors were installed inside and outside the house (

Table 1. List of instruments and sensors and their installation locations for real-time measurement.

Measurement	Instrument or Sensor	Location
PM10 concentration	TEOM × 3	Fan 4 and mobile lab roof
Fan rotational speed	Hall Effect sensor × 46	Fans 1–46
House differential pressure	differential pressure transmitter × 3	East and west side walls
RH/T	RH/T transmitter × 2	Fans 5 and 28
Temperature	Type T thermocouple × 5	House, raceway, and TEOM pump chamber
Activity	Passive Infrared Detector × 2	Fans 5 and 28
Lighting	Light sensor × 1	Fan 5
Manure belt operation	Current switch × 2	Manure belt control box
Fan airflow speed	Impeller anemometer × 2	Fans 3 and 4
Weather condition	Wireless weather station × 1	Rooftop of the processing plant
Total	67

). The PM₁₀ concentrations were measured continuously with three TEOM (Tapered Element Oscillating Microbalance) instruments (model 1400a Ambient Particulate Monitor, ThermoFisher Scientific, Franklin, MA, USA). Its mass transducer minimum detection limit is 0.01 µg. The instrument has a precision of ±5.0 µg m⁻³ for 10 min averaged data and ±1.5 µg m⁻³ for 1 h averages [36]. The TEOM monitor is designated by the U.S. EPA as an Automated Equivalent Method (EPA Designation No. EQPM-1090-079) [37] for PM₁₀ and used extensively in state and national PM monitoring networks.

A TEOM instrument consists of two major units, a sensor unit and a control unit, which are connected with cables and tubing. The TEOM has a built-in barometric pressure transducer to monitor atmospheric pressure in the control unit. Three TEOMs were used in this study. They were factory-calibrated, and the end of the study occurred prior to the calibration expiration date for each unit. The control units of the three TEOMs were all set up inside the mobile lab.

2.2.3. Data Acquisition and Control

An on-site computer system (OSCS) acquired data from all on-line instruments and sensors at 1 Hz. It also monitored and controlled the temperature inside the raceway, which was heated to protect the air inside the TEOM tubing from condensation. The OSCS consisted of a personal computer, custom software AirDAC (Version 4.3), and data acquisition and control hardware. The AirDAC converted signals from instruments and sensors to engineering units, performed data pre-processing, averaged the data over 15 s and 1 min intervals, and saved the mean values in two separate data files [17].

2.3. PM₁₀ Sampling and Concentration Measurement

2.3.1. Exhaust Air PM₁₀ Concentration

Concentrations of PM₁₀ in the house exhaust air was the most important variable for PM₁₀ emission monitoring. To ensure data quality and data completeness, the house PM₁₀ concentrations in the exhaust air were measured directly and continuously with two TEOMs close to the air inlet of fan 4, which was one of the always-on Stage 1 minimum winter ventilation fans (Figure 2).

The sensor units of the two TEOMs were housed in two wooden protection boxes with the sensor heads standing above the boxes. The centers of the two TEOM inlets were placed 0.19 m from the wall and 1.0 m from each other. The two TEOM sensor units, including the inlets, were named TEOM-N, which was north of the other and closer to fan 4, and TEOM-S, which was south of the other. The center of the TEOM-N inlet to the edge of the fan 4 shutter opening was 1 m (Figure 2).

2.3.2. Outdoor Air PM₁₀ Concentration

The outdoor PM₁₀ was sampled on the roof of the mobile lab (Figure 1). The TEOM for outdoor measurement was named TEOM-O, of which the sensor unit was inside the mobile lab below the sensor inlet. The center of the TEOM inlet in this study was 11.90 m from the west sidewall of the house.

2.4. Other Measurements

2.4.1. Ventilation Rate Measurement

To obtain reliable ventilation rate measurements, a combination of different methods was applied:

• All 46 fans were monitored individually and continuously for their blade rotational speeds using NPN Hall Effect sensors (Model NJK-5002C, made in Taiwan).
• Differential static pressures across the house’s west and east side walls, in which the ventilation fans were installed, were continuously measured using differential pressure sensors (Model 260, Setra Systems, Boxborough, MA, USA).
• A portable Wall Fan Tester was designed and built to determine the on-site fan ventilation rates under different fan rotational speeds and differential static pressures.
• Exhaust air speeds were measured continuously at two selected fans (fans 3 and 4) using 18 cm vane axial anemometer (Model 27106T, R.M. Young Co., Traverse City, MI, USA) for real-time fan operation check as a quality assurance measure.

Using the results of fan tests with the Wall Fan Tester, a fan model was developed to convert the measured fan rotational speeds and differential pressures to fan airflow rates for all the fans every minute. More details about the ventilation measurement in this study are described in [34].

2.4.2. Weather Measurement

A weather station (Model 6152 Wireless Vantage Pro2, Davis Instruments, Hayward, CA, USA) was installed on the roof of the egg processing plant of the farm. The measured weather variables included wind direction and speed, outdoor temperature and relative humidity (T and RH), solar radiation, and precipitation. The weather data were transferred wirelessly and in real time to a Vantage Pro2 Console (Davis Instruments), which was installed in the mobile lab, and saved every minute to the data acquisition and control computer with the same timestamps as all the other instruments and sensors connected to the OSCS.

2.4.3. Indoor Temperature and Relative Humidity Measurement

Indoor T and RH were measured close to fan 5 on the west side and close to fan 28 on the east side of the house. Two wall-mount T/RH Transmitters (Model RS-WS-120-2, Jianda Renke, Shangdong, China) were used (Table 1). The transmitters have probe operating T from −40 °C to +80 °C (accuracy of ±0.5 °C at 25 °C) and RH from 0% to 100% (accuracy of ±3% RH from 5% to 95% at 25 °C).

Additionally, two copper–constantan (Type T) thermocouples were installed side by side with the T/RH transmitters at fans 5 and 28 as backups. Two other Type T thermocouples were used to sense temperatures in the heated raceway. One more thermocouple was installed in the mobile lab chamber, where three TEOM pumps were housed. The temperatures measured with the thermocouples were for quality control of the monitoring system, not for emission calculation.

2.4.4. Hen Activity, House Lighting, and Manure Belt Operation Monitoring

For quality control of the on-site monitoring, several other variables were continuously monitored during the entire study period (Table 1). They included animal/worker activities at selected locations, house lighting schedule, and manure belt operation.

Two Passive Infrared Detectors (PIDs, Model SRN-2000 Detector, Visonic Inc., Bloomfield, CT, USA) were installed on the east and west walls next to the T/RH transmitters. The PIDs detected close-by activities of hens and workers and output semi-quantitative signals. The high intensities of hen and worker activities are usually correlated to high PM concentrations [17].

The house lighting was measured with a custom-made lighting sensor that contained a Photosensitive Sensor Module for Arduino (ASIN: B01N1FKS4L, Amazon.com, Seattle, WA, USA). The light sensor was installed next to the PIDs and close to fan 5. The sensor output was a continuous analog voltage, corresponding to light intensity, but was mainly used to monitor the on/off status of the indoor lighting.

Two current switches (P/N: CR9380-NPN, CR Magnetics, St. Louis, MO, USA) were connected to the unused contacts of the manure belt motor relays in conjunction with digital inputs of the data acquisition system. The digital signal outputs of the current switches indicated the moving or stopping status of the belts and were used for monitoring quality control purposes.

2.5. Data Processing, Analysis, and Interpretation

2.5.1. Data Processing Procedure and Software

Data processing was performed using custom-developed software CAPECAB (Version 18.09) that followed the NAEMS Standard Operating Procedure (SOP) B6 [38]. The CAPECAB software program allows immediate access to the data to visualize and inspect the data. It can calculate emission values every minute, every hour, every day, or other time intervals.

On-line measurement data from AirDAC were imported to CAPECAB [39]. Raw data that were averaged in 1 min intervals were first checked graphically in CAPECAB along with the experimental notes. Any abnormal data were flagged and did not participate in follow-up data processing and analysis. Weather data from the wireless weather station and hen data provided by the producer were also imported into CAPECAB. The timestamps of all the data were lined up in the software.

2.5.2. PM₁₀ Emission Calculation

Calculation of PM₁₀ emission rate followed the general equation for net aerial pollutant emissions from animal buildings, i.e., only counting the pollutant generated in the buildings by subtracting the outdoor pollutant concentrations from the house exhaust concentrations, as shown in Equation (1).

$$E=Q\times \left(C_E-C_O\right)$$

(1)

where E is net aerial pollutant emission rate, mass time⁻¹; Q is ventilation rate or air exchange rate, volume time⁻¹; and C_E and C_O are aerial pollutant concentrations at the air outlet and inlet of the building, respectively, mass volume⁻¹.

Emissions of PM₁₀ presented in this paper were converted to PM₁₀ concentrations at standard conditions of 20 °C and 1 atm [Equation (2)], following an SOP for the NAEMS (Section 6.7.2 in [40]).

$$E_{PM}=\left\{\left[Q\times P_E\times {\displaystyle \frac{(273+20)}{(273+T_E)}}\right]\times \left(C_E-C_O\right)\right\}$$

(2)

where E_PM is net PM emission rate of the house, mass time⁻¹; Q is house ventilation rate at T_E, volume time⁻¹; P_E is pressure of fan exhaust air, atm; C_E is PM₁₀ concentration of fan exhaust air, mass volume⁻¹; C_O is outdoor PM₁₀ concentration, mass volume⁻¹; T_E is temperature of exhaust air, °C.

The average of PM₁₀ concentrations measured with TEOM-N and TEOM-S was used as C_E. The outdoor ambient PM₁₀ concentration measured with TEOM-O was used as C_O. The house ventilation rate Q was the sum of ventilation rates from all 46 fans. The pressure of fan exhaust air P_E (at the TEOM-N and TEOM-S heads) was obtained by adding the outdoor atmospheric pressure measured with the weather station and the differential pressure across the west wall. The outdoor atmospheric pressure was a positive value from the weather station. The differential pressure was recorded as a negative value because the house was slightly depressurized.

Three units of daily mean emissions, i.e., total house emission, emission per hen, and emission per animal unit (AU = 500 kg animal mass), were used to express the house PM₁₀ emission rates. Because the house went through a molting period, during which egg production was drastically reduced from 25 July 2021 until the gradual recovery one month later, the emission rate per egg was not considered representative and is not used in this paper.

2.5.3. Data Analysis and Emission Assessment

The main statistics analyzed in this report include maximum, minimum, mean, standard deviation, and 95% confidence interval (c.i.) for hourly and daily means. Daily mean values of relevant variables are plotted to demonstrate variation patterns and trends. Correlations and curve fitting were employed to analyze relationships between different variables.

The number of valid data hours for each variable was the sum of hours, each of which had >75% (45 min) of valid data. Similarly, the number of valid data days for each variable was the sum of days, each of which had >18 h (>75%) of valid data. Average hourly mean (AHM) is the average of valid hourly means. Average daily mean (ADM) is the average of valid daily means. Because the numbers of valid hourly data and valid daily data are different, the standard deviations (Std) and 95% c.i. between the AHM and ADM are not the same, although they are calculated from the same dataset.

Emission rates for the egg farm were assessed based on the emission rates per day per hen from the house using Equation (3). Emission rates per day per hen were also assessed by comparing with reported data in the U.S. and other countries.

$$E_{PM\_Farm}=E_{PM\_House}\times {\displaystyle \frac{He{n}_{Farm}}{He{n}_{House}}}$$

(3)

where E_{PM_Farm} and E_{PM_House} is net PM₁₀ emission rate from the egg farm and the house, mass time⁻¹, respectively, and Hen_Farm and Hen_House are hen inventory on the egg farm and in the house, n, respectively.

3. Results and Discussion

3.1. Data Completeness

Compared with the monitoring project goal of obtaining ≥75% valid hourly means, i.e., ≥3285 valid hourly means, set forth for this study, the actual data completeness was well above this targeted level. The total monitoring time from 16 March through 15 September 2021, was 4416 h or 184 days. The numbers of valid hours for the main measurement and calculation variables ranged from 97.9% (TEOM-N PM₁₀ concentrations) and 100% (weather conditions). The completeness of hourly mean and daily mean PM₁₀ emission rates from the house were 98.7% and 98.4%, respectively.

As a comparison, in a 6-month study from 1 August 2004 to 31 January 2005 [41], the PM₁₀ data collected from the control house with 154,004–172,522 hens had 81% data completeness (the number of days with over 70% valid data, instead of 75% in this study). Another measurement in the NAEMS, the PM₁₀ emission data completeness, was even lower, at <50% in one of the manure belt houses, because of some technical issues during the study [17]. High data completeness rates can reduce biases that often result from missing important data, hence satisfying one of the critical requirements to ensure high research quality (

Table 2. Data completeness in the 6-month (4416 h or 184 days) study from 16 March through 15 September 2021.

Data Type	Hourly Means		Daily Means
	Valid Hours, h	Completeness, %	Valid Days, d	Completeness, %
Weather data	4416	100.0	184	100.0
Ventilation rate	4369	98.9	181	98.4
PM10-N STP	4322	97.9	179	97.3
PM10-S STP	4371	99.0	182	98.9
PM10-O STP	4379	99.2	182	98.9
PM10 emission from the house	4357	98.7	181	98.4

Note: The PM₁₀ emissions were calculated from PM₁₀ concentrations under standard temperature and pressure conditions.

).

3.2. Outdoor Environmental Conditions

3.2.1. Outdoor Temperature and Relative Humidity

The minimum hourly and daily mean outdoor temperatures during this study were −6.2 °C on 2 April from 6:00 to 7:00 a.m. and −0.8 °C on 1 April (

Table 3. Hourly and daily mean weather conditions, measured with the on-site weather station.

Statistics	Outdoor T, °C	Outdoor RH, %	Solar Radiation, W m−2	Wind Speed, km h−1	Atmospheric Pressure, mb
Hourly means
Range	−6.2 to 34.2	20.2 to 95.7	0.0 to 991.0	0.0 to 58.1	975 to 1014
AHM ± Std	18.5 ± 7.3	70.8 ± 17.2	210.5 ± 272.3	10.4 ± 6.7	995 ± 6
AHM ± 95% c.i.	18.5 ± 0.2	70.8 ± 0.5	210.5 ± 8	10.4 ± 0.2	995 ± 0
Daily means
Range	−0.8 to 27.0	34.4 to 91.9	16.6 to 336.5	2.4 to 25.2	977 to 1011
ADM ± Std	18.5 ± 6.4	70.8 ± 12.3	210.5 ± 70.0	10.4 ± 4.8	995 ± 5
ADM ± 95% c.i.	18.5 ± 0.9	70.8 ± 1.8	210.5 ± 10.1	10.4 ± 0.7	995 ± 1

Note: AHM = average hourly mean; ADM = average daily mean; Std = standard deviation; c.i. = confidence interval.

). The maximum hourly and daily mean outdoor temperatures were 34.2 °C on June 29 from 3:00 to 4:00 p.m. and 27.0 °C on 27 June 2021. The relative humidity during the study ranged from 20.2 to 95.7 in hourly means and 34.4 to 91.9 in daily means.

The average AHM temperature and relative humidity during the study were 18.5 ± 7.3 °C and 70.8 ± 12.3%, respectively. A general increasing trend in both outdoor temperature and relative humidity was shown during the study (Figure 3, left). These temperatures and relative humidity represented the local weather conditions from March to September.

Outdoor temperature plays a critical role in pollutant concentrations in animal buildings because it has direct effects on indoor temperature and building ventilation rate. Higher outdoor temperature results in a higher building ventilation rate, which effectively dilutes indoor airborne pollutants and reduces their concentration. The effect of outdoor relative humidity on pollutant emissions is more complex.

3.2.2. Solar Radiation, Wind Speed, and Atmospheric Pressure

The maximum hourly mean solar radiation was 991 W m⁻². The daily mean solar radiation ranged from 16.6 to 336.5 W m⁻². The average daily mean solar radiation was 210.5 ± 70.0 W m⁻² (Table 3 and Figure 3, right). Solar radiation affected indoor temperatures, which in turn affected house ventilation rate and indoor PM₁₀ concentrations.

The recorded maximum hourly mean wind speed was 58.1 km h⁻¹. The daily mean wind speeds ranged from 2.4 to 25.2 km h⁻¹. The average daily mean wind speed was 10.4 ± 4.8 km h⁻¹. The hourly mean atmospheric pressure fluctuated from 975 to 1014 mb during the study. The AHM and ADM were 995 ± 6 and 995 ± 1 mb, respectively. Wind speeds were found to have considerably affected house differential pressures and efficiencies of ventilation fans in this monitoring project [34].

3.3. Indoor Environmental Condition

3.3.1. Indoor Temperature and Relative Humidity

The indoor temperatures and relative humidity measured in the west side (at fan 5, near the PM₁₀ sampling location) and the east side (at fan 28) showed slight differences between the two locations. The west side had 1.7 °C and 0.1 °C lower minimum and maximum hourly mean temperatures, respectively, than the east side (

Table 4. Hourly and daily mean indoor temperature (T), relative humidity (RH), and differential static pressure (dP).

	Fan 5 T, °C	Fan 28 T, °C	Fan 5 RH, %	Fan 28 RH, %	East dP, Pa	West dP, Pa
Hourly means
Range	18.7 to 33.5	20.4 to 33.6	23.1 to 83.6	25.4 to 82.0	−64.3 to 3.1	−64.2 to −1.5
AHM ± Std	25.1 ± 2.5	25.9 ± 2.5	57.0 ± 11.3	56.8 ± 11	−17.7 ± 5.8	−18.6 ± 6.1
AHM ± 95% c.i.	25.1 ± 0.1	25.9 ± 0.1	57.0 ± 0.3	56.8 ± 0.3	−17.7 ± 0.2	−18.6 ± 0.2
Daily means
Range	20.1 to 29.7	21.3 to 30.5	36.9 to 77.9	37.0 to 77.9	−26.9 to −3.4	−35.6 to −4.5
ADM ± Std	25.1 ± 2.1	25.9 ± 2.0	57.0 ± 9.3	56.8 ± 9.3	−17.7 ± 4.0	−18.6 ± 4.0
ADM ± 95% c.i.	25.1 ± 0.3	25.9 ± 0.3	57.0 ± 1.3	56.8 ± 1.3	−17.7 ± 0.6	−18.6 ± 0.6

Note: AHM = average hourly mean; ADM = average daily mean; Std = standard deviation; c.i. = confidence interval.

). The ADM temperatures were 25.1 ± 2.1 and 25.9 ± 2.0 °C for the west and east sides, respectively. The daily mean temperatures generally increased from mid-March to mid-September (Figure 4).

Smaller differences in indoor relative humidity between the west and east sides were recorded during the study. The west side (fan 5) had 2.3% lower and 1.6% higher hourly minimum and maximum relative humidities than the east side (fan 28). The AHM and ADM relative humidities were both 57.0% in the west side and 56.8% in the east side (Table 4). Relative humidity also generally increased from mid-March to mid-September (Figure 4).

The differential pressures across the west and east sidewalls, primarily created by the operation of the wall ventilation fans and affected by house air inlet opening and wind velocity, were similar to each other. Both the AHM and ADM differential pressures were −17.7 Pa across the east wall and 18.6 Pa across the west wall (Table 4). A detailed analysis of house differential pressures has been presented in [34].

3.3.2. House Ventilation Rate

The technical approaches used in this study for ventilation determination considerably improved data quality, as detailed in [34]. The hourly mean house ventilation rates, which included all 46 fans, ranged from 1342 to 22,436 m³ min⁻¹. The ADM house ventilation rate was 10,803 ± 5906 m³ min⁻¹. The ADM ventilation rate per hen was 0.078 m³ min⁻¹.

On days of very low outdoor temperatures, the east wall ventilation fans (fans 24–46) generated lower ventilation rates than the west wall fans (fans 1–23) because of the configuration of fan stages. The two stage 1 fans (fans 4 and 19) were both on the west sidewall. Therefore, when only one stage of fans was running, there were no fans on the east wall running.

Over the six months of study, the daily mean house ventilation rates increased from March to the end of August, followed by a decrease to the end of the study. However, analysis of the data revealed that the house ventilation was well controlled based on outdoor temperature over the entire study period [34].

3.3.3. House Lighting

The continuous measurement of house lighting intensities revealed a dynamic lighting schedule that had 14 changes during the six-month study. Major changes in lighting schedule started four days before the artificial molting began and continued through the end of the study. During this period, the daily lighting durations ranged from 10 to 24 h. A sudden decrease in lighting duration to 10 h per day occurred on 20 July and continued until 9 August. The lighting duration then gradually increased from 11 August, returning to 16 h with two-period lighting by 16 September. More details about the dynamic lighting in the house and its impact on PM₁₀ concentrations and emissions are presented in [42].

3.4. PM₁₀ Concentration

3.4.1. PM₁₀ Concentration at Ambient and STP Conditions

Because PM₁₀ concentrations were converted to STP conditions before performing emission calculations, results of both units of PM₁₀ concentrations, i.e., at ambient conditions and at STP conditions, are available, as presented in

Table 5. Hourly and daily mean PM₁₀ concentrations measured with north (N), south (S), and ambient TEOMs.

Statistics	Non-Converted PM10 Concentrations, µg m−3				PM10 Concentrations at STP, µg m−3
	North	South	N, S Mean	Ambient	North	South	N, S Mean	Ambient
Hourly means
Range	−137 to 3856	−110 to 4081	−123 to 3968	−11 to 250	−131 to 7833	−105 to 4004	−118 to 3893	−11 to 243
AHM ± Std	255 ± 283	226 ± 230	243 ± 257	33 ± 26	246 ± 273	218 ± 222	235 ± 248	33 ± 25
AHM ± 95% c.i.	255 ± 8	226 ± 7	243 ± 8	33 ± 1	246 ± 8	218 ± 7	235 ± 7	33 ± 1
Daily means
Range	69 to 1134	57 to 913	65 to 977	6 to 86	66 to 1094	54 to 873	62 to 942	6 to 85
ADM ± Std	253 ± 179	226 ± 148	244 ± 169	33 ± 16	244 ± 173	218 ± 142	236 ± 162	33 ± 15
ADM ± 95% c.i.	253 ± 26	226 ± 21	244 ± 25	33 ± 2	244 ± 25	218 ± 21	236 ± 24	33 ± 2

Note: AHM = average hourly mean; ADM = average daily mean; Std = standard deviation; c.i. = confidence interval; STP = standard temperature and pressure.

. However, only the concentrations at STP are discussed in this paper, unless otherwise explained.

3.4.2. Indoor and Outdoor PM₁₀ Concentration

After converting to STP conditions, the hourly mean PM₁₀ concentrations ranged from −131 to 7833 µg m⁻³, from 105 to 4004 µg m⁻³, and from −11 to 243 µg m⁻³ in the TEOM-N, TEOM-S, and TEOM-O measurements, respectively. The negative PM₁₀ concentrations were inherent with the instrument. Adsorption and desorption of moisture and semi-volatile species may cause positive or negative artifacts in TEOM PM mass measurement [43]. These negative values did not show in the daily mean PM₁₀ concentrations, which averaged 244 ± 173 µg m⁻³, 218 ± 142 µg m⁻³, and 33 ± 15 µg m⁻³ at TEOM-N, TEOM-S, and TEOM-O at STP conditions, respectively (Table 5).

3.4.3. Temporal Variations in PM₁₀ Concentration

Temporal variations in PM₁₀ concentrations at the fan exhaust were clearly shown over the 6 months of monitoring and can be divided into two periods. The daily mean PM₁₀ concentrations exhibited a slowly decreasing trend from 15 March to 15 August and an increase from 16 August to the end of the study (Figure 5, left, TEOM-N and TEOM-S). Increasing PM₁₀ concentrations were seen from 15 August to the end of the study. The outdoor PM₁₀ concentrations were relatively stable (Figure 5, left, TEOM-O, y = 0.1315x − 5799.3, R² = 0.2163). Molting had a remarkable impact on the PM₁₀ concentrations in the house (Figure 5, left).

3.4.4. PM₁₀ Concentration Difference Between TEOM-N and TEOM-S

Daily mean comparison of PM₁₀ concentrations for the 179 days (valid data days of TEOM-N) showed 244 ± 173 µg m⁻³ and 219 ± 143 µg m⁻³ at TEOM-N and TEOM-S, respectively. Although the responses of the two TEOMs to PM₁₀ concentrations were linearly correlated (R² = 0.96, Figure 5 Right), the ADM concentration for TEOM-S was 10.2% lower than TEOM-N. This difference was attributed to the general technical issue of variations among different instruments.

Because all the TEOMs in this study had been manufacturer-refurbished and calibrated and no operational problems were identified during the study, it was impossible to identify the measurement of which TEOM was closer to the “true” PM₁₀ concentrations. Therefore, the average PM₁₀ concentrations of TEOM-N and TEOM-S was used for PM₁₀ emission calculation, except for three days (13–15 September), when TEOM-N was not functioning and only the data from TEOM-S were available.

3.4.5. Pegged PM₁₀ Concentration

The highest outdoor 1 min PM₁₀ concentrations was 866 µg m⁻³ measured with TEOM-O at 10:46 22 June 2021. The highest recorded indoor 1 min PM₁₀ concentrations with TEOM-N and TEOM-S was 9000 µg m⁻³. This was because the measurement range of the two TEOMs was set from −1000 to 9000 µg m⁻³. During the routine house cleaning time using a power dust blower near the indoor TEOMs by the farm staff, the airborne PM₁₀ concentrations sometimes went extremely high and exceeded 9000 µg m⁻³. The total time that the TEOMs were pegged were 62 and 22 min for TEOM-N and TEOM-S, respectively (

Table 6. Duration of pegged PM₁₀ measurement (>9000 µg m⁻³) data at the two indoor TEOMs.

Date	TEOM-N, min	TEOM-S, min
7 March 2021	2
31 March 2021		10
7 April 2021	3
15 April 2021	8
19 April 2021	6
12 May 2021	4
31 May 2021		6
31 May 2021	4
21 August 2021		6
25 August 2021	4
6 September 2021	21
8 September 2021	9
10 September 2021	1
Total	62	22

).

The pegged TEOM issue has occurred in other measurement studies in laying hen houses, including the NAEMS. When the two TEOMs were pegged in this study, the maximum concentration recorded in the data acquisition computer was 9000 µg m⁻³. The true concentrations during that time are unknown. However, because of the short percentages of time compared with the total monitoring time of 184 days (0.02% for TEOM-N and 0.008% for TEOM-S), the potential errors it could introduce into the ADM PM₁₀ concentrations and emissions were minimal.

3.5. PM₁₀ Emission Rate

3.5.1. PM₁₀ Emission from the House

The hourly mean emissions from the house ranged from −3.7 to 40.4 kg d⁻¹ (

Table 7. Hourly and daily mean net PM₁₀ emissions from the house in three different units and from the farm using two different calculation methods.

Statistics	House, kg d−1	House, mg d−1 AU−1	House, mg d−1 hd−1	Farm, kg d−1 (1)	Farm, kg d−1 (2)
Hourly mean
Range	−3.7 to 40.4	−9175 to 94,892	−26.5 to 292.7	−54.2 to 518.8	−48.3 to 534.5
AHM ± Std	2.6 ± 3.2	6265 ± 7721	18.7 ± 23.3	35.2 ± 45.4	34.5 ± 44.9
AHM ± 95% c.i.	2.6 ± 0.1	6265 ± 229	18.7 ± 0.7	35.2 ± 1.3	34.5 ± 1.3
Daily mean
Range	0.4 to 10.2	998 to 24,784	3.1 to 74.9	5.4 to 148.2	5.6 to 147.9
ADM ± Std	2.6 ± 2.1	6319 ± 5015	18.9 ± 15.3	35.6 ± 31.1	34.8 ± 30.9
ADM ± 95% c.i.	2.6 ± 0.3	6319 ± 731	18.9 ± 2.2	35.6 ± 4.5	34.8 ± 4.5

Note: AHM = average hourly mean; ADM = average daily mean; Std = standard deviation; c.i. = confidence interval. ⁽¹⁾ Calculated based on emissions per AU from the house; ⁽²⁾ Calculated based on emissions per hen from the house.

). The negative PM₁₀ emissions in hourly mean emissions were related to the corresponding PM₁₀ concentrations. These ranges represented large variations in hourly mean emissions over the 6 months. When daily mean emissions were calculated, these variations were reduced. The daily mean emissions from the house ranged from 0.4 to 10.2 kg d⁻¹.

After averaging the entire duration of the study, the AHM and ADM PM₁₀ emissions from the house were 2.6 ± 3.2 (2.6 ± 0.1 AHM ± 95% c.i.) and 2.6 ± 2.1 (2.6 ± 0.3 ADM ± 95% c.i.) kg d⁻¹, respectively. The ADM emissions per AU (animal unit = 500 kg animal body weight) and per hen were 6319 ± 5015 mg d⁻¹ AU⁻¹ and 18.9 ± 15.3 mg d⁻¹ hd⁻¹, respectively.

3.5.2. Temporal Variation in PM₁₀ Emission

The patterns of PM₁₀ emissions from the house were similar to those of PM₁₀ concentrations (Figure 5, left). Three periods with different emission qualities and patterns were demonstrated (

Table 8. Comparison of selected statistics during three monitoring periods at the house.

Statistics	Three Emission Periods
	16 March–27 April	28 April–13 August	14 August–15 September
Outdoor temperature, °C	10.66	20.56	22.24
Outdoor relative humidity, %	62.35	71.65	78.42
Indoor temperature, °C	22.96	25.72	25.98
Indoor relative humidity, %	50.19	57.42	64.25
House ventilation, m3 min−1	4494	12,267	14,366
PM10-N concentration, µg m−3	200.4	196.9	468.4
PM10-S concentration, µg m−3	188.5	173.4	397.1
PM10-O concentration, µg m−3	24.23	32.24	44.79
House emission rate, kg d−1	1.08	2.07	6.20
Emission rate per hen, mg d−1 hen−1	7.70	15.00	45.50

Note: Temperature and relative humidity were measured at fan 5.

). The lowest emission period was from 16 March to 27 April, during which the ADM was 1.08 kg d⁻¹ from the entire house. The second period started on 28 April. When the molting in the house began in July, there was a mild decrease in PM₁₀ emissions. This trend continued until 13 August, the end of the second emission period (Table 8). The ADM emission rate was 2.07 kg d⁻¹ during this period. The temporal variation patterns in these periods were similar and were not substantial.

However, starting from 14 August until the end of the study, a large change in house emissions was seen. The emission rates increased remarkably, and the ADM during this period was 6.20 kg d⁻¹. This transition was observed during and beyond the molting period and was directly linked to the dynamic lighting schedule change that affected hen activities [42]. These results demonstrated that molting had a substantial impact on PM₁₀ emissions from the house.

3.5.3. Effect of Concentration and Ventilation on PM₁₀ Emission

Analysis of the relevant on-line and real-time data in the study revealed that the house exhaust PM₁₀ concentrations were similar between the first two periods but were higher in the third period (Table 8 and Figure 5, left). The house ventilation rates increased from the first period to the third period until the end of August, followed by a period of lower ventilation rates in September. Therefore, the changes in PM₁₀ emissions could be directly related to both the variations in PM₁₀ concentrations and ventilation rates. The changes in house ventilation rates could be logically explained by the variations related to outdoor temperatures during the study [34].

3.5.4. Effects of Molting on the Egg Farm PM₁₀ Emission

Among the fourteen hen houses on the egg farm, only three went through molting in 2021. In addition to the house under study, molting in a second house occurred from 3 May to 24 June and in a third house from 7 June to 29 July. Therefore, molting only occurred in 21% of the houses in 2021, and the molting duration of 52 days was only about 14% of the days in a year. This is equivalent to an occurrence of about 3% of the total egg production days on the farm in 2021.

3.5.5. PM₁₀ Emission from the Egg Farm

Using PM₁₀ emissions per AU and per hen from the house to calculate the total emissions from the egg farm resulted in two slightly different ADM values. The ADMs based on emissions per AU and per hen were 35.6 ± 31.1 kg d⁻¹ (35.6 ± 4.5 ADM ± 95% c.i.) and 34.8 ± 30.9 (34.8 ± 4.5 ADM ± 95% c.i.) kg d⁻¹ from the egg farm, respectively (Table 7). The annual emissions from the egg farm, based on these ADM values, were 14.3 tons per year (12,994 kg year⁻¹), calculated per AU, and 14.0 tons per year (12,702 kg year⁻¹), calculated per hen. Both annual emission rates were below the emission threshold of ≥250 tons per year for Type A sources and ≥100 tons per year for Type B sources—primary PM₁₀ [44], p. 178.

3.6. Comparison with the Reported PM₁₀ Concentration and Emission

Concentrations and emissions of PM₁₀ from five manure belt laying hen houses with long-term and continuous monitoring were found in the available literature (

Table 9. Comparison of PM₁₀ concentrations and emissions from long-term studies in commercial manure belt laying hen houses using the TEOM for PM concentration measurement reported in the literature.

Location, House, Duration	PM10 Concentration, µg m−3	PM10 Emission (mg d−1 hen−1)	Reference
Ohio, one house, 6 months	256 ± 108	12.4 ± 13.3	[45]
Indiana, house A, 2 years	415 ± 428	12.4 ± 19.3	[17,18]
Indiana, house B, 2 years	761 ± 661	25.2 ± 33.3	[17,18]
Ohio, house 1, 1 year	497 ± 232	18.0 ± 9.6	[16]
Ohio, house 2, 1 year	522 ± 168	17.8 ± 9.6	[16]
Ohio, one house, 6 months	236 ± 162	18.9 ± 2.2	This study

). These studies were all conducted with a similar laying hen house air quality measurement protocol using the TEOM for PM₁₀ concentration measurement.

Compared with literature data of five manure belt laying hen houses, the ADM PM₁₀ concentration of 236 ± 162 µg m⁻³ (average of TEOM-N and TEOM-S) at STP was very close to 256 ± 108 µg m⁻³ measured in the same house in 2004. However, it was much lower than the five other houses in Indiana and Ohio and was only about 57% and 31% of House A and House B at the Indiana site, respectively [18].

The study reported by Purdue University [45], conducted in the same house as this study, used the same monitoring protocol from August 2004 to February 2005. It was under daily mean outdoor temperatures from −13.7 to 26.2 °C and the ADM outdoor temperature of 10.6 ± 9.8 °C, lower than this study’s temperature of 18.5 ± 6.4 °C (Table 3). The ADM indoor temperature was 22.9 ± 0.4 °C, also lower than this study’s temperature (25.1 ± 2.1 °C at Fan 5 and 25.9 ± 2.0 °C at Fan 28, Table 4). Because of the lower outdoor temperatures, the ADM ventilation rate per hen was 0.046 m³ min⁻¹, only about 60% of this study’s rate. Under these weather and house environment conditions, the emission rate reported by Purdue University [45] was 12.4 ± 13.3 mg d⁻¹ hen⁻¹.

Compared with the results in [45], the emission rate of 18.9 ± 2.2 mg d⁻¹ hen⁻¹ during the entire six months in this study was 52.4% higher. However, the ADM emission rate in this study during the first two emission periods from 16 March to 13 August was only 12.9 ± 5.8 mg d⁻¹ hen⁻¹ (Table 8), very close to the six months study reported in [45]. Evidently, the higher emissions during and after the molting in the third emission period (Figure 5 and Table 8) contributed to the higher emission rate compared with the 2004–2005 study.

Nevertheless, despite the molting, compared with the four other manure belt laying hen houses reported in Indiana and Ohio, the ADM emission rate in this study was still lower than one house at 25.2 ± 33.3 mg d⁻¹ hen⁻¹ (Table 9). The overall PM₁₀ emissions during the entire study period were still within the range reported in the literature.

4. Conclusions

The following conclusions were drawn from this study:

(1)

High-level quality assurance and quality control played a major role in achieving >98% PM₁₀ emission data completeness.

(2)

Long-term and continuous measurement are necessary to cover seasonal and diurnal variations in PM₁₀ concentrations and emissions and to characterize air pollution at animal buildings.

(3)

Measurement results from the two side-by-side TEOMs demonstrated that different instruments can introduce variations in measurement results.

(4)

During the routine house cleaning time using a power dust blower near the indoor TEOMs by the farm staff, the airborne PM₁₀ concentrations sometimes went extremely high and exceeded 9000 µg m⁻³.

(5)

The ADM in-house PM₁₀ concentration at STP, averaged from the two TEOMs, was 236 µg m⁻³, close to the 256 µg m⁻³ measured in the same house in 2004–2005. The ADM outdoor PM₁₀ concentration at STP was 33 µg m⁻³.

(6)

The net ADM PM₁₀ emissions from the house in three different units were 2.6 kg d⁻¹ house⁻¹, 6319 mg d⁻¹ AU⁻¹ and 18.9 mg d⁻¹ hen⁻¹. The ADM PM₁₀ emission of 18.9 mg d⁻¹ hen⁻¹ was within the range reported in the literature, which was between 12.4 and 25.2 mg d⁻¹ hen⁻¹ for manure belt laying hen houses.

(7)

Comparison with the PM₁₀ emissions from the same house in 2004–2005 revealed that artificial molting was related to a remarkable increase in PM₁₀ concentrations and emissions.

(8)

Based on the emission per hen per day from the house, the estimated ADM PM₁₀ emission from the entire egg farm was 35.6 kg d⁻¹. The annual emission from the egg farm was 12,702 kg year⁻¹ (14.0 tons per year), below the emission threshold of ≥250 tons per year for Type A Sources and ≥100 tons per year for Type B Sources for Primary PM₁₀.