Assessment of long-term gas sampling design at two commercial manure-belt layer barns

Understanding temporal and spatial variations of aerial pollutant concentrations is important for designing air quality monitoring systems. In long-term and continuous air quality monitoring in large livestock and poultry barns, these systems usually use location-shared analyzers and sensors and can only sample air at limited number of locations. To assess the validity of the gas sampling design at a commercial layer farm, a new methodology was developed to map pollutant gas concentrations using portable sensors under steady-state or quasi-steady-state barn conditions. Three assessment tests were conducted from December 2008 to February 2009 in two manure-belt layer barns. Each barn was 140.2 m long and 19.5 m wide and had 250,000 birds. Each test included four measurements of ammonia and carbon dioxide concentrations at 20 locations that covered all operating fans, including six of the fans used in the long-term sampling that represented three zones along the lengths of the barns, to generate data for complete-barn monitoring. To simulate the long-term monitoring, gas concentrations from the six long-term sampling locations were extracted from the 20 assessment locations. Statistical analyses were performed to test the variances (F-test) and sample means (t test) between the 6- and 20-sample data. The study clearly demonstrated ammonia and carbon dioxide concentration gradients that were characterized by increasing concentrations from the west to east ends of the barns following the under-cage manure-belt travel direction. Mean concentrations increased from 7.1 to 47.7 parts per million (ppm) for ammonia and from 2303 to 3454 ppm for carbon dioxide from the west to east of the barns. Variations of mean gas concentrations were much less apparent between the south and north sides of the barns, because they were 21.2 and 20.9 ppm for ammonia and 2979 and 2951 ppm for carbon dioxide, respectively. The null hypotheses that the variances and means between the 6- and 20-sample data were equal at alpha = 0.05 (P > 0.05) were accepted for both gases. The results proved that the long-term gas sampling design was valid in this instance and suggested that the gas sampling design in these two barns was one of the best on the basis of available long-term monitoring instrumentation at reasonable cost.