Enhancing indoor air quality in dairy farms via automatic air monitoring and mitigation strategies.

The identification and characterization of the current indoor air quality will be critical for the overall sustainability and competitiveness of the US dairy industry as it relates to the health of animals and agricultural communities. This project addresses the urgent need for quantifying the dynamics of air components identified as air pollutants including ammonia, carbon monoxide, carbon dioxide, sulfur dioxide, methane, nitrogen dioxide, and ozone (EPA, 2023a; EPA 2023b; WHO, 2021) and for developing practical strategies to improve air quality in dairy farms. The long-term goal of this project is to enhance the indoor air quality of commercial dairy farms via the utilization of real-time air quality measurements that can support management decisions in cow housing and milking parlor spaces. This research will open possibilities to establish air quality standards linked to the overall impact of dairy operations on environmental, social, and economic sustainability. Moreover, this research will provide easily available information about the appropriate localization of air monitoring systems in dairy farms, air quality dynamics, and strategies to improve air quality regarding cleaning and ventilation interventions. The specific objectives that will support the achievement of the long-term goal of this research are:

Research objectives
Objective 1: To determine the accuracy and precision of automated air quality monitoring systems and their appropriate localization in naturally and cross-ventilated cow housing barns and milking parlors.
Activity 1.1: Air quality monitoring sensors (IEQMax®, USA), consisting of an air analytic platform with individual sensors measuring the air components specified in Table 1, will be used to assess indoor air quality in cow housing barns and milking parlors in commercial dairy farms. These air monitoring systems allow passive flow of the air inside of the analytic platform and deliver real-time air component concentration every minute.

To test sensor accuracy, precision, and appropriate allocation, two commercial dairy farms located in Northern Colorado will be enrolled in a prospective study. The first study farm (Farm A, Figure 1) has one naturally ventilated barn and a robotic parlor with six milking robotic units. Farm A milks a total of 380 lactating Holstein and Brown Swiss cows housed in one indoor free-stall barn provided with water beds, automatic manure scrapping robots, and two manure collection pits (Figure 1). The second study farm (Farm B, Figure 1) milks a total of 6,000 lactating Holstein cows housed in three free-stall cross-ventilated barns provided with sand bedding and mechanically operated manure vacuum and scrapping systems. For this project, one cow housing barn will be monitored for air component dynamics in Farm B.
At the beginning of the project, air monitoring sensor accuracy will be compared between the readings of one sensor installed at the center of the cow housing barns in Farms A and B and air samples collected using 3 L air sampling bags (Tedlar, Dupont, USA) and an air pump. Air samples will be submitted for analysis at the Powerhouse Energy Institute at Colorado State University. Air samples will be collected at 0.6, 6, 13, 26, 52, and 104 feet from the sensor to estimate reading accuracy and range during a one-day sampling campaign performed in years 1 and 2 of this project. As the milking parlor is a separate room in Farm B (Figure 1), air samples will be collected in the milking parlor at the same distances to evaluate air sensor accuracy. The air sampling times will be recorded, and Pearson’s correlation coefficients will be calculated between sensor reads and laboratory results.

Activity 1.2: Precision between sensor measurements installed at different horizontal and vertical distances will be estimated. This procedure will be performed as follows: 1) Horizontal comparisons of sensor readings between sensors installed near the barn gates and at 25 and 50% of the total length of the barns (Figure 1); 2) Vertical comparisons of sensor readings between sensors installed at 5 and 13 feet from the ground at the center and near the barn gates. Concordance correlation coefficients will be calculated using data collected for one week during the cold and hot seasons. In addition to the data provided by the air monitoring sensors, daily outdoor air quality data provided by the EPA (EPA, 2023c) including carbon monoxide, nitrogen dioxide, ozone, and PM 2.5 and 10.0 will be retrieved during the life of this project to contrast indoor and outdoor air quality.
The outputs of activities 1.1 and 1.2 will provide appropriate localization of air monitoring systems regarding data reliability and the range of measurements to estimate the number of sensors needed to cover barns and milking parlors.

Objective 2: To identify the dynamics of air components and group scale emissions in indoor dairy spaces associated with dairy farm operations.
Activity 2.1: Three air monitoring sensors will be installed at the center and near the barn gates in each study farm (Figure 1). In farm B, one additional sensor will be deployed at the center of the rotary milking machine. The concentrations of air components considered in this study (Table 1) will be measured every minute. This information will be automatically stored in a server cloud (Grafana, Amazon Web Services, USA) and will be downloaded weekly and safely stored in Colorado State University’s drives. The observation period will consider 12 months of continuous monitoring after the completion of accuracy and precision analysis. As this data is highly condensed, air component measurement will be averaged by hour of the day and by date. This approach will allow us to investigate daily patterns of air component concentrations that might be related to farm operations such as animal feeding, vehicle movement in the barns, manure scrapping, and animal movements. In addition to air component concentrations, group scale emission will be estimated. These metrics will combine information provided by the air monitoring sensors and farm information including the number of animals in the cow housing barns, group dry matter intake (DMI), and daily milk yield (kg/day). Thus, daily group scale emission estimates including production (unit/day), emission yield (unit/kg of DMI), and emission intensity (unit/kg of milk) will be calculated. In addition, wind speed will be continuously monitored using anemometers (RXW-WCF-900, Onset Computer Corporation, MA) installed in both study dairies to estimate the emission rates of the air components evaluated in this project. We will generate reports of air component concentrations and group scale emissions dynamics considering descriptive statistics, regression analysis, and graphics that will allow us to examine the time of the day, locations in the dairy, and periods of the year in which exposure of the air components considered in this study reach maximum levels.

Objective 3: To design and implement practical mitigation strategies for improving air quality in naturally and cross-ventilated cow housing barns and milking parlors.
Our preliminary data (not published) indicates that specific farm activities such as feeding time, curtain activation (Farm A), the start and finish of the milking shifts (Farm B), and the use of fans and sprinklers may influence air component concentration dynamics in dairies. Using this evidence, we will develop mitigation strategies aiming to reduce air pollutant concentrations and group-scale emissions. These strategies will be focused on ventilation and cleaning interventions. We hypothesize that incorporating data-driven decisions for pen cleaning, bedding changes, ventilation, manure management optimization, and refining feeding practices will enhance air quality in dairy facilities.
Activity 3.1: Farm A is naturally ventilated with automated curtains in the North and South walls of the barn (Figure 1). The baseline concentration of air components will be measured with the curtains down during feeding, scrapping, and during non-operation times (no feeding, no vehicle movement). Then, air sensor measurements will be collected with the curtains rolled up at 25, 50, and 100% of their opening capacity. Levels of air component concentrations will be compared between the curtains position strategies. In the ongoing assessment in Farm B, we have observed a notable increase in PM 2.5 during morning hours (maximum 4.8 ug/m3). Moreover, we have detected stable ammonia and carbon dioxide concentrations (maximum 9.6 and 562 ppm, respectively) during night hours followed by a notable decline at the beginning of the morning shifts (minimum 3.22 and 441 ppm, respectively), suggesting a potential link between ventilation system operation and air quality fluctuations as forced ventilation is activated during day hours. In consequence, in Farm B, we will assess the effect of forced ventilation on reducing PM2.5. ammonia, and other air components (Table 1) in the cow housing barn by performing a controlled intervention on the fan ventilation systems. First, baseline measurements of the air components (Table 1) will be assessed during feeding and sand bedding replacement times with the ventilation fans turned off. Second, measurements from the air sensors will be collected during the same period as the ventilation fans are turned on. Finally, the levels of air components will be compared between the two ventilation approaches. For interventions in Farms A and B, the time in which concentration drops to steady minimum levels will be assessed to outline timing recommendations for curtain and fan systems activation to enhance air quality, which might be translated into more cost-efficient use of ventilation to maintain air quality and appropriate environmental conditions in dairy farms. These assessments will be performed thrice during the Fall, Winter, Spring, and Summer.
Activity 3.2: As manure accumulates in the cow circulation alleys, pen scrapping procedures must be continuously performed to reduce the accumulation of ammonia and ensure proper cow hygiene and milk quality. Farm A uses automatic alley-scrapping robots that collect and dump manure in the collection pits at the center of the cow housing barn (Figure 1, Farm A). We will assess the effect of increasing pen frequency on reducing the concentration of ammonia and other air components. Baseline air component concentrations will be measured with the current alley scrapping schedule. Then, we will subsequently assess scrapping frequency increases of 1, 2, and 3 times per day for three consecutive days. Multiple comparisons of air component concentrations will be performed to determine the effect of scrapping frequency on air quality in naturally ventilated barns.
Activity 3.3: In the milking parlor of Farm B, we will assess the use of water sprinklers on the overall air quality. Our ongoing research has shown that an increasing volume of PM 2.5 (maximum 17.6 ug/m3) follows the same patterns as THI increments, which are controlled using water sprinklers and fans. Thus, the effects of different levels of water sprinkler power (0, 50, 100%) on the concentration of PM 2.5 and other air components will be tested during the hot season. Adjusting sprinkler usage based on weather conditions and pollutant levels will lead to significant water and energy savings while maintaining or even enhancing air quality benefits. Multiple comparisons between air component concentrations by sprinkler use strategies will be performed.
Activity 3.4: Statistical analysis. This activity encompasses data preparation and statistical analyses (SAS 9.4, SAS Institute Inc., Cary NC) for activities 1.1 to 3.3. Separated mixed models will be built by the study farm to estimate the least squared means (LSM) of the air component dynamics and group scale emissions values measured in this project (Table 1). The LSM will be calculated by hour of the day, date, and season including interaction terms in the model. In addition, the effect of farm activities and interventions specified in activities 2.2 to 3.3 on the air components concentrations and group scale emissions will be estimated using two-way ANOVA. Statistical significance will be determined at a P-value <0.05 and covariates with a P-value ≤0.1 will be retained in the models for confounder control.