Know more

Background and challenge

Dairy farms are complex systems subject to the vagaries of changing markets, climate, landscapes and public expectations. A better understanding of what drives their performance is essential for the long-term sustainability of the dairy sector. Today’s newest technologies can contribute to this goal by providing automated real-time monitoring of the production, reproduction, health and welfare of every animal in a herd.

Performance in a dairy business has many drivers, from management practices (choice of feed, breeding and rearing system, genetic selection) and the performance of individual animals (depending on the apportionment of resources towards different functions such as growth, milk production, and reproduction) to environmental influences. The inSiliCow project sets out to create a simulation tool that can take the combination of these different factors into account, operationalising the concept of the digital twin through its application to a real dairy farm.

The project will enable the creation of a virtual dairy farm, through which the different available strategies to manage individuals, the herd and the farming system can be tested. It will be possible to use this virtual farm as an innovative decision-making tool to improve the economic, social and environmental performance of real farms.

Goals and methodology

InSiliCow is an individual-based multiscale dairy cow simulator. The simulator is built around the integration into a single tool of multiple ‘layers’, ranging from the conceptual representation of a cow’s physiology to a computer architecture that allows individual models to be managed within a virtual herd:

1. A virtual representation of the physiology of a dairy cow based on the description of the dynamic interactions between the priorities of each of the animal’s various biological functions (e.g. growth, reproduction, maintenance and ageing).
2. A mechanistic model of energy flows and transactions that determine the dynamics of a dairy cow’s performance (e.g. ingestion, constitution and use of body reserves, gestation and lactation).
3. A mechanistic model, in which priority dynamics and energy flows are coupled, taking the form of a virtual cow that simulates the phenotype of a real cow from birth to death according to its genotype.
4. A mechanistic model of a cow’s reproduction, making it possible to simulate the sequence of reproductive cycles and the generation of offspring in the form of new individual cow models.
5. A model of a farming system that allows rules to be specified for the management of the individuals in a herd (e.g. feeding, insemination, selection, culling, renewal, etc.)
6. A computer architecture that allows the management of individual models of cows within a virtual herd according to the selected farming system and individual performance.

This four-year project will design and produce an operational digital twinning tool (simulation and coupling of observed and simulated data).

Through partnerships with INRAE’s experimental farms and those of its international partners (see project partner list) a large dataset will be created at herd scale.

The digital twin will be used to address new scientific questions that enable better understanding and modelling of a cow’s metabolism, and to develop herd management strategies to improve the health and welfare of farmed animals.

Contact - coordination :

• Olivier Martin, UMR MoSAR
• Charlotte Gaillard, UMR PEGASE

Units involved and partners

The project brings together 12 units from 4 INRAE scientific divisions (PHASE, GA, MATHNUM and ECOSOCIO), including three experimental units (UEP, Herbipole and PAO), two higher education establishments (AgroParisTech and VetAgro Sup), the IDELE, and three international partners: KU Leuven (Belgium) LUKE (Finland) and Aarhus University (Denmark).

INRAE participating units

Partners

Partenaires internationaux