Conference Agenda

Accurate ammonia (NH₃) emission measurements in naturally ventilated (NV) livestock houses remain challenging. A direct method using airflow rate measurements with ultrasonic anemometers recently gained attention as an alternative to the commonly applied tracer gas ratio (TGR) method, which relies solely on gas concentrations. Both methods require accurate incoming barn air concentrations to correct outgoing concentration measurements.

This study presents emission measurements with both methods in four NV dairy barns in Flanders with gabled roofs and large northeast and southwest oriented ventilation openings. For the direct method, airflow rates were determined every ten minutes, while outside gas concentrations were determined at an hourly rate at both sides as well as inside the barn. The side with incoming concentrations was identified using three approaches:

(1) incident wind angle from meteorological data,

(2) the airflow direction measured by anemometers in the openings (direct method), and

(3) the side with lowest outside NH₃ or CO₂ concentrations (TGR method).

First results revealed that parallel wind incidence angle frequently produced variable airflow directions during hourly concentration measurements. Under those conditions, the lowest outside concentrations of NH₃ or CO₂ sometimes occurred at opposite sides. This study will evaluate implications on NH₃ emission measurements for both methods.

Methane (CH₄) emissions from pig houses primarily originate from manure storage and are an environmental concern. Developing emission prediction models helps us understand the sensitivity of individual influencing factors so that targeted mitigation strategies could be suggested. The CH₄ emission models developed for pig houses equipped with side-wall jet ventilation and pit ventilation predicted emissions reasonably well (R² = 0.87), but their prediction accuracy for pig houses with diffuse ceiling ventilation, widely used in Denmark, was reduced (R² = 0.66). This study developed CH₄ emission models specifically for pig houses installed with diffuse ceiling ventilation with and without pit ventilation. The models integrate sub-models for the mass transfer coefficient derived from CFD simulations and slurry surface concentration models trained and validated using experimental measurements from a section containing two pig pens. Results show that the CH₄ mass transfer coefficient is positively correlated with ventilation rate and negatively correlated with pig body mass. The mass transfer coefficient without pit ventilation is approximately one-third of that with pit ventilation. The surface concentration model is explicit and physically interpretable, achieving good performance with pit ventilation (R² = 0.85). Without pit ventilation, the model performs well from November to January (R² = 0.82).

Water evaporation from poultry litter plays a crucial role in moisture content and emission processes in poultry barns. We studied the effects of induced variations in environmental and litter conditions on water evaporation from laying hen litter using a dynamic chamber system. Litter samples varied in initial dry matter (DM) (70%, 80%, 90%), structural condition (homogeneous and non-homogeneous), and surface refreshing type (refreshing and non-refreshing), and were evaluated across six environmental blocks (10, 20, 30 °C all with 50 and 80% RH).

The results showed that the evaporation rates increased when samples were placed at higher temperatures with lower ambient humidity (20-30 °C at 50% RH), regardless of their initial DM. Across treatments, the evaporation flux presented a positive relationship with the water vapour pressure difference between litter and air. With that, we found water activity of 70-90% DM samples were in the range of 0.9-0.65. Moisture flux of samples were approximately linear constant over four days. Under specific conditions with 90% DM and 80% RH, litter samples absorbed water. For non-homogeneous samples, variation over time was substantially higher than in homogeneous treatments.

This finding will later be used to perform computational fluid dynamics (CFD), gaining accuracy in the computations.

Accurate measurement methods are needed to evaluate the effect of mitigation strategies on ammonia emissions from livestock houses. Dynamic chambers (DC) have been widely used for this purpose as they enable the monitoring of emissions under controlled, adjustable airflow rates. However, internal airflow patterns and ammonia transport dynamics are not yet well quantified. Therefore, this study aimed to investigate the effect of varying suction rates and ammonia concentrations on internal airflow dynamics.

For this, Computational Fluid Dynamics simulations of internal airflow dynamics were conducted in DC over a range of ammonia concentrations (25-100 ppm) and air suction rates (0.1-10 L.min^-1). Strong relations were found between suction rate, ammonia concentration distribution and airflow patterns. Stagnant and active mixing zones were distinguished using a critical diffusion-convection threshold velocity of 1.4×10⁻⁴ m s^-1. At low suction rates (≤0.5 L.min^-1), gas transport appeared to be essentially dominated by molecular diffusion, resulting in the presence of stagnant zones, delaying gas removal. Conversely, suction rates above 1 L.min^-1 enhanced mixing, resulting in uniform gas dispersion.

These findings highlight the significant effect of suction rate on gas detection, measurement variation and accuracy. Determining it correctly is therefore essential to get accurate and reliable measurements from DC.

Livestock farming is a major source of atmospheric ammonia (NH₃), requiring continuous monitoring for mitigation. However, direct measurements in naturally ventilated systems are expensive and require complex instrumentation, resulting in limited large-scale adoption. Therefore, machine learning (ML) models that use routinely recorded farm information as input parameters for estimating barn air characteristics are increasingly being studied. This study was designed to evaluate the reliability of machine-learning models for estimating hourly indoor NH₃ concentrations. Measurements collected from April 2025 to January 2026 (5,225 hourly observations) in a naturally ventilated dairy barn were combined with 18 predictors describing environmental conditions, management activities, herd characteristics, and milk production indicators. Under randomly mixed training and testing conditions, CatBoost achieved the highest accuracy (R² ≈ 0.84) among the four tree-based ensemble models (Random Forest, HistGradientBoosting, XGBoost, and CatBoost). Moreover, feature analysis showed that management activities, herd production state, and indoor climate conditions had interacting effects on model estimation. However, the estimation performance declined substantially under time-aware validation. Blocked validation produced R² values of 0.22–0.57, while a lag^-1 persistence benchmark achieved R² > 0.80, indicating strong temporal autocorrelation. These findings emphasize time-aware evaluation for machine-learning-based livestock pollution monitoring, accounting for temporal dependence and seasonal variability.

Quantifying methane (CH₄) and ammonia (NH₃) emissions from naturally ventilated dairy barns (NVDBs) is technically challenging and resource-intensive. Ventilation rates (Q) end emissions are commonly estimated using the indirect CO₂ balance method. Yet its inter-laboratory precision, accounting for both the laboratory performing the measurements and the instruments used within each laboratory, has not been systematically assessed. In this study, a field-scale round-robin campaign involving five independent laboratories was conducted to evaluate methodological reproducibility. Each laboratory simultaneously measured indoor and outdoor gas concentrations at a naturally ventilated dairy barn at the Leibniz-InnoHof near Potsdam, Germany, using five independent sampling systems equipped with either cavity ring-down spectroscopy (CRDS) or Fourier transform infrared spectroscopy (FTIR). Emissions were estimated using the CO2-balance method as described in the VERA protocol. Measurements were done over a whole week, resulting in ~1100 observations per analyzer. Although analyzer-specific differences in absolute concentrations were observed, the resulting ventilation and emission estimates showed strong agreement across laboratories (PCC > 0.96). The study demonstrates a relative precision better than ±8% (95% CI) for Q, CH₄, and NH₃. These results confirm the robustness of the CO₂ balance method for applications such as national emission inventories and the validation of emission mitigation strategies.

Manure management is a major source of methane emissions, and mitigation instruments for existing livestock facilities are urgently needed in order to achieve short-term methane emission reduction.
In this work, we present the combination of SDS (sodium dodecyl sulfate), a biodegradable surfactant that inhibits methane production, and weekly discharge of slurry from pig production as a promising strategy to reduce me-thane emissions across the manure management chain.
SDS for slurry treatment has been developed through lab experiments simulating slurry dynamics including discharge from housing to external storage. Based on these experiments, the most cost-efficient dosing strategy was identified and implemented in an experimental production-scale pig house. To minimize SDS consumption and take advantage of the effect of weekly discharge in itself, SDS was added to the residual slurry (after discharge) by using a custom-built dosing system that enable controlled dosing and distribution of SDS beneath the slatted floor. Results from lab and pilot-scale experiments show that the combined technology can reduce the in-house slur-ry emissions by ~75%, while the overall emission reduction (in-house + ex-ternal storage) is in the range of 40-50% %. The results are supported by model simulations. Potential implementation barriers will be discussed.