Many efforts have been made to bring agroindustry into a more technological and digital environment. The so-called "Agro 4.0" is speeding up with the collection of a huge amount of data, the intensive use of automation, and geospatial mapping through drones, for example. Most of these practices can substantially increase the productivity and quality of the food produced.
One of them is mathematical optimization. Some of its benefits have already been mentioned in a previous insight. Here we will talk about a specific application: defining animal diet formulation. The industry has already been optimizing this issue with very simple linear models for decades. The question is: is it possible to improve such decision-making by enlarging the scope of decisions, taking advantage of today's technologies and knowledge?
The diet problem is a classic optimization problem and it was one of the first problems studied in the 1930s and 1940s. The motivation came from the Army's desire to minimize the cost of feeding the soldiers in the field, while still providing them with a healthy diet.
As with most classic optimization problems, its application can be adapted. We can apply this to the dairy cows’ feeding diet: feed costs account for the greatest portion of the variable costs of producing milk (it can reach 60% of total production cost). Therefore, proper planning of the nutritional management of the herd avoids unnecessary expenses and favors profits in the activity.
Usually, the first step is to divide the cattle into groups with similar characteristics, such as genetic, age, and lactation. Based on these characteristics, with the support of veterinary and zootechnical doctors, the minimal and maximal requirements of each nutrient can be identified. These requirements become constraints in the model. To satisfy these constraints, we need to know the nutritional information of each ingredient available, which will then be multiplied by the amount consumed (our variables).
The consumption of the available food has also an availability constraint and will incur production or purchase costs. These costs will be in the model objective function and will be minimized. This is the essence of this diet formulation optimization model, and can already result in cost reduction.
However, many other aspects can be considered, such as predictive models to better prevent metabolic diseases (i.e. sub-clinical rumen acidosis) and more consistent control of milk quality (i.e. milk fatty acid profile). Even the environmental impact can be included in an optimization model. Another improvement is to consider productivity as a variable, based on the nutritional input of the diet, and turn the model into profit optimization, taking into account the revenue from milk production.
The complexity level of such a model can vary depending on the farm/company maturity, and it can evolve with time, adding new features as results appear. What is certain is that this is a great opportunity to reduce costs and increase profit using mathematical modeling, with no impact on animal well-being!
Author: Cassiano Lima - Senior Consultant at Cassotis Consulting
Co-author: Emmanuel Marchal - Managing Partner at Cassotis Consulting