Adsa Logo White Adsa Title White

A dynamic model to predict herd consistency and milk deliveries of a dairy herd.

A. S. Atzori



Join A. S. Atzori on this page for a live text chat!
6:00 PM - 8:00 PM GMT


A dynamic model to predict herd consistency and milk deliveries of a dairy herd.
A. S. Atzori*1, A. Gallo2. 1Dipartimento di Agraria, University of Sassari Sassari, Italy, 2Department of Animal Science, Food and Nutrition (DIANA), Universit� Cattolica del Sacro Cuore Piacenza, Italy.

Dynamic modeling applied to dairy farming allows improving the farm decision making process trough prediction of future performances. This work aimed to define a preliminary model structure able to minimize the system complexity and to reach a satisfactory prediction dynamics of a given Mediterranean dairy herd and milk deliveries. The modeling process had the objective of farm output. The modeling process followed a System Dynamics approach and included: i) a preliminary farm survey to gather technical info and data; ii) the development of a Stock and Flow model on Vensim (Ventana Inc.), to mathematically and graphically simulate the dynamic connections among variables; iii) the model evaluation against farm records. The survey provided aggregated monthly records, from Jan 2015 to Dec 2018 (4 years), of consistency, feed supply, milk deliveries, reproduction indexes, mortality and health of herd categories. Average farm characteristics in the same period consisted of 1154 � 58 milking cows, 205 � 35.7 dry cows, 1375 � 45.2 heifers whereas dry matter intake (DMI) and milk yield were on average equal to 24.2 � 2.4 and 34.0 � 3.4 kg/d per head, respectively. The model included 25 variables distributed on a closed aging chain of 4 stocks (replacement heifers and open, pregnant and dry cows) and the respective flow rates of calving, breeding, drying off and culling. Initial settings were based on farm values observed in Jan 2015. Two farm inputs (monthly average of consumed feed from lactating cows and conception rate) were included as exogenous variables resulting of fundamental relevance in model predictions. The simulation showed a good accuracy in predicting the oscillating seasonal pattern of historical farm records of milk deliveries. Cow consistency and milk deliveries were predicted with a RMSPE of 3.5% and 7.2% of observed values (being 57% and 68% due to random variability of data), with high accuracy (Cb = 0.90 and 0.97), with good precision (r2 = 0.68 and 0.72), and with a satisfactory concordance correlation coefficient (0.74 and 0.82), respectively. Model updates should focus on wheater effects on intake and reproduction at farm level.

Keywords: system dynamics, dairy management, weather.