Optimisation and validation of the intermittent measurement method to determine ammonia emissions from livestock buildings

Abstract

In previous studies, a method to reduce the high cost of continuous ammonia emission measurements by using the expensive equipment in several buildings for short periods was developed. Further optimisation and validation of the method of “intermittent measurements” to estimate the ammonia yearly emission from pig houses, based on a limited number of measuring days, are described. The method is tested and improved with new data from three different pig houses. The method of “intermittent measurements” models the relationship between ammonia emission and easily measurable variables such as in- and outdoor temperatures, ventilation rate and weight of the animals. The parameters of the emission model are determined for an individual farm in a time variant manner. They are based on frequent ammonia measurements in combination with the easily measured variables from a limited number of days over a certain time interval. For all other days throughout the year, the emission of ammonia is not measured, but, by using the model, it is calculated from the easily measured variables. In previous studies, the parameters in the emission model for a specific building were calculated only once a year with the data of all selected measuring days throughout the year. In this study, the ammonia emission was modelled in a more precise, stepwise manner such that model parameters were calculated over shorter time periods, such as 70 days. This resulted in an optimised procedure in which, for fattening pigs, each fattening period was modelled in two parts: the first part from day 1 to day 70 and the second part from day 70 to the end of the fattening period. The adapted method was tested on data from seven fattening periods originating from three different pig houses. The accuracy of the method was obtained by evaluating the difference between the ammonia emission calculated from the model with the measured total ammonia emission, sampled every 12 min over a whole fattening period. With 4 measuring days per fattening period, a maximum model error of less than 10% was achieved for all datasets; while previous method with fixed model parameters throughout the year on the new validation datasets resulted sometimes in errors above 25%.