Composting animal manure with natural aeration is a low-cost and low-energy process that can improve nitrogen recycling in millions of farms world-wide. Modelling can decrease the cost of choosing the best options for solid manure management in order to decrease the risk of loss of fertilizer value and ammonia emission. Semi-empirical models are suitable, considering the scarce data available in farm situations. Eleven static piles of pig or poultry manure were monitored to identify the main processes governing nitrogen transformations and losses. A new model was implemented to represent these processes in a pile considered as homogeneous. The model is based on four modules: biodegradation, nitrogen transformations and volatilization, thermal exchanges, and free air space evolution. When necessary, the parameters were calibrated with the data set. The results showed that microbial growth could reduce ammonia volatilization. Greatest nitrogen conservation is achieved when microbial growth was limited by nitrogen availability.