Prediction of Temperature and Humidity in a Naturally Ventilated Pig Building

A mathematical description is given of heat- and moisture-transfer processes in a pig house, for 880 weaned piglets, equipped with natural ventilation, floor heating and local air cooling by means of additional humidification. Dynamic discrete auto-regressive moving average (ARMA) models, describing the air temperature and humidity inside the building and continuous models as transfer functions are derived. Experiments were performed to identify the structure and coefficients of the discrete ARMA models, using recursive prediction error (maximum-likelihood and least-squares) methods. The input and output data for the identification experiments were collected from experiments carried out over three seasons, viz., winter, spring and summer. The structure of the discrete models was established and it was found that the dynamic properties of the heat- and moisture-transfer processes could be described using first-order dynamic models. The model orders were determined by Akaike's test criterion. The identified models were validated by tests to check whether the residuals consisted of white noise. The models were also tested with data sets which were not used in the identification process and their predictions agreed well with measured values. The mean value of the differences between measured data and simulated values for the temperature model was 1·12°C and 1·02 g/kg for the humidity model. Static models describing the air temperature and humidity inside the livestock building in the steady state were also derived. The use of these models offer a quick engineering method for determination of the required supplemental water flow rate for humidifying the indoor air in summer and the required supplemental heat in winter.