CAP-DAIRY: computer aided planning of dairy farms

Abstract

A linear programming (LP) model has been developed (CAP-DAIRY) to describe the grass utilisation and feeding system on a dairy farm. It links several components of the system and optimizes the system as a whole. The model links a grass utilisation model, a feed ration model and a novel model which relates on-going milk yield to on-going feeding level. The main feature of the model is the approach adopted to relate feeding levels, milk yield and weight changes. When cows are fed more than they require for maintenance and the current level of milk yield, the excess energy becomes increased bodyweight and cause an increase in milk yield. When fed less than they require, they mobilize reserves into energy for milk production and lose weight and tend to reduce milk yield. At the start of the lactation some weight loss is tolerated. This is treated in the model as a requirements for up to 0.5 kg/day weight loss in addition to maintenance so that a lower weight loss is the increase case. A linear mathematical model that represents this mechanism was developed and incorporated into the LP and fitted to data which changed the level of feeding of dairy cows during the lactation. This gives a greater flexibility to the LP and allows the model to determine optimal feeding levels at all stages of the lactation and as a consequence optimal milk yields and optimal stages for weight changes, which vary depending on calving date and feed availability The grass utilisation model permits the successive utilisation of grass for grazing or silage making. Grazing can take place two, three or four weeks after the previous use and silage making five, six or seven weeks after the previous use. To allow for the effect of silage making on regrowth, use after this is delayed by one week. Data on energy and dry matter yields at any time is required and the model determines the optimum schedule of use and frequency. Silage is made in a number of separate silos but the feeding-out energy value makes the model non-linear. This is solved by using a recursive approach in which the initially unknown feeding value is calculated from successive solutions and the model reoptimized to convergence. The feed ration model determines the amount of grass, silage and concentrates required based on the maximum dry matter intake, which is a function of yield and the energy required for maintenance, milk yield and any weight change. The model could be easily extended to also use protein given suitable data. The LP determines the optimal land use for forage and cash crops, calving pattern and feeding strategy according to specific farm conditions such as farm area, milk quota and availability of forage maize. Several scenarios were studied and the effects of changes of different parameters analysed. Results indicated that net margins increased with maize crop areas and gave higher optimum milk yields replacing concentrates up to an optimum area of maize. The seasonality of milk prices affected particularly calving pattern and milk yield and the results suggested they led to more even milk production due to encouraging Autumn calving. Results also showed that the optimal feeding levels is different for cows calving in different periods of the year resulting in different weight change pattern and milk yields. Spring calving cows lost more weight than cows calving in any other period, but regained the weight lost quickly. They also produced the lowest level of milk. Autumns calving cows had the highest milk yield and the lowest weight losses, although a longer period to regain that weight was optimal. Summer calving cows produced slightly less milk and lost slightly more weight than Autumn cows. Another important aspect that results showed was the influence that maize silage has on farm decisions. The larger the maize crop area the higher the marginal price of milk quota, showing that milk quota constraint was more severe for those cases and consequently higher prices for extra milk quota could be paid. CAP-DAIRY is suitable for analysing the impact that changes such as milk prices, fertilizer prices or concentrates prices would cause on the optimal plans. The model is also helpful to evaluate research priorities by analysing the effects caused by biological and technical changes such as grass varieties and machinery.