Estimating efficiency in automatic milking systems

Abstract

Milking data of 34 single automatic milking system (AMS) units on 29 Galician dairy farms were analyzed to determine the system capacity in each farm under actual working conditions. Number of cows, milk yield, milkings per cow per day, actual milking time, rejected milking time, cleaning time, and machine downtime were used to determine the number of cows milked per AMS unit to obtain the optimal values of milkings per cow and milk production. Multiple linear regression data analysis was used to model the linear relationship between the dependent variable, milk yield per AMS per year, and the predictor variables: number of cows per AMS, milkings per cow per day, milk flow rate, and rejections per AMS per year. An AMS unit milked 52.7±9.0 cows daily at 2.69±0.28 milkings per cow, with a total milking downtime of 1,947±978h/yr and a milk yield of 549,734±126,432kg/yr. The predictor variables cow and milk flow rate had a greater level of influence on the milk yield per AMS than milkings per cow and rejections, and explained the 87% of the variation. The AMS in Galician dairy farms could facilitate an increase of 16±8.5 cows per AMS without impairing milking performance; in this way, the quantity of milk obtained per robot annually could be increased (185,460±137,460kg). This would make it possible to recoup the cost of the system earlier. In the present situation, the daily milking throughput could be maximized at 2.4 to 2.6 milkings per cow.