DIGITAL TWIN PROTOTYPES IN FLOW-THROUGH SYSTEMS FOR FINFISH

Introduction

The virtual, digital counterpart of a physical object referred as digital twin derives from the Internet of Things (IoT) and involves real time acquisition and processing of large data sets. A fully implemented system ultimately enables real-time and remote management, as well as the reproduction of real or forecasted scenarios. Despite such potential, the adoption of digital twin features by smaller enterprises, including by aquaculture SMEs, has been comparatively slow (Uhlemann et al., 2017).

Under the emerging framework of Precision Fish Farming, we set up digital twin prototypes for land-based farms of Rainbow trout (Oncorhynchus mykiss), European seabass (Dicentrarchus labrax) and Gilthead seabream (Sparus aurata), with the aim of supporting producers in optimizing feeding practices and oxygen supply with respect to 1) growth performances; 2) fish welfare, and 3) environmental loads. The digital twins were conceptualized targeting rearing cycles at Preore Farm (Trentino-Alto Adige, Italy), for trout, and Vigneto Farm (Tuscany, Italy), for seabass and seabream. The twins rely on integrated mathematical models which are fed with farm data sets and simulate several dynamic processes, allowing the estimation of key parameters such as feed digestibility, fish appetite, ammonia excretion rate, fish size distribution and dissolved oxygen consumption.

Methodology

For the trout cultivation system, the envisaged digital twin block is to be implemented to the liquid oxygen storage and supply system, whereas for the seabass and seabream plant, the focus is on optimizing feeding. These twin blocks will be interconnected to bioenergetic, population management and dissolved oxygen dynamic models, schematized in Figure 1.

The bioenergetic model (Brigolin et al., 2014) simulates the fish growth and metabolism. The forcing variables input to this model are water temperature, feeding rate, feed composition, initial average fish weight and population dynamics variables, which are associated with the population management model. In addition, husbandry parameters are input, including species-specific optimal and lethal extreme temperatures and energy consumed by respiration. The resulted calculated fish biomass is then input into a dissolved oxygen (DO) transport model (Royer et al., 2021; Lima et al., 2021), with the aim to assess the rapidly changing DO concentration in the cultivation ponds.

Results and discussion

The results demonstrated the potential of the models to provide the digital twins with short-term response capabilities, specifically intra-hour adjusted oxygen supply and daily adjusted feeding. The bioenergetic model provided robust estimates for the evolution of individual fish and total biomass weights, based on a single initial value for fish weight. With these values as inputs, the feeding table in the envisaged digital twin is automatically corrected on daily basis. At present, in contrast, these corrections tend to occur at longer time windows, and weight estimates are provided based on sampling. It is noticeable that the uncertainties in sampled fish weights may be significant, as indicated by the fact that farmers may back calculate fish weights once the biomass has been fully harvested.

The assessment of oxygen dynamics in the trout cultivation system considered scenarios based on current practices at the farm, fish welfare, the quality of the water discharged from the cultivation tanks and temperature changes. The results for DO concentration in a spatial-temporal frame obtained from the model indicate that the consumption of liquid oxygen can be reduced significantly with a real-time control.

Acknowledgements

The research leading to these results has received funding from the European Union’s HORIZON 2020 Framework Programme under GRANT AGREEMENT NO. 773330 and 862658.

References

Brigolin, D., Meccia, V.L., Venier, C., Tomassetti, P., Porrello, S., Pastres, R., 2014. Modelling biogeochemical fluxes across a Mediterranean fish cage farm. Aquaculture Environment Interactions 5, 71-88.

Lima, A. C., Royer, E., Pastres, R., 2021. Dissolved oxygen consumption and intra-tank distribution of trout biomass. AE2020, European Aquaculture Society, 137-138.

Royer, E., Faccenda, F., Pastres, R., 2021. Estimating oxygen consumption of rainbow trout (Oncorhynchus mykiss) in a raceway: A Precision Fish Farming approach. Aquacultural Engineering, 92, 102141.

Uhlemann, T. H. J., Schock, C., Lehmann, C., Freiberger, S., Steinhilper, R., 2017. The digital twin: Demonstrating the potential of real time data acquisition in production systems. Procedia Manufacturing 9, 113–120.