Plenary Lecture

Prof. Antonis Kokossis

AI-assisted innovations in the twinning of bioreactors and the synthesis of industrial biocatalysts

The presentation outlines trends and challenges in the development of digital twins, highlighting opportunities and the scope to define and refine the digital twin concept in the context of specific applications. The presentation further focuses on bioreactors and specific incentives to promote the digital twin concept in industrial biotechnology, an application domain very fragmented but also extremely rich in data and knowledge. Bioreactor digital twins are discussed in general terms and then, in relation to Bioindustry 4.0, a collective research effort to formulate generic and systematic approaches that are suitable to explore twinning for use in a wide range of biotechnology services. The presentation shares recent results that explain the bioreactor digital twin concept, the potential of technology to shadow and twin real-life processes, as well as the scope to the digital replica for optimization and control.

Building on this foundation, the presentation explains recent results demonstrating the use of Physics-Informed Neural Networks (PINNs) to create high-fidelity digital twins by integrating

first-principle dynamic models (ODEs) with sparse experimental data. In a pilot study involving recombinant nanobody production with E. coli, PINNs were applied to both batch and fed-batch bioreactor phases. Digital shadows proved successful to accurately replicate system dynamics and enable real-time state estimation and kinetic parameter inference. Even with a limited dataset, digital twins remarkably provide excellent short-term predictions, significantly outperforming conventional regression approaches. The framework supports continuous model refinement as new measurements become available, making it extremely attractive for adaptive process control and dynamic optimization. In most fed-batch applications, digital twins demonstrated strong adaptability by dynamically handling discontinuous inlet flow changes — a major challenge in bioprocess modeling. Despite the complexity introduced by variable feed rates, PINN-based models consistently delivered robust short- and long-term predictions. Additionally, kinetic parameters can be estimated without prior knowledge, highlighting an ability to generalize across phases. The presentation concludes with the integration of process twins with biochemical and metabolomics, supporting strain engineering and retrosynthesis as curated kinetic parameters can be combined with digital threads pointing to options for overexpressing or downregulating paths of the metabolic networks for better reaction and process efficiency.

Keynote Lectures

Prof. Mariano Martín Martín

Multiscale Aproaches to Develop a Biocircular Economy

Biowaste is one of the major issues in our society. The volume, its distributed availability and composition presents important management challenges. However, biowaste valorization and exploitation are essential towards energy and food security as well as for social development of the rural areas. Only through a multiscale analysis, from the selection of the chemical/product to obtain to the strategic investment plan it is possible to come up with guidelines towards deploying a biocircular economy.

The methodoly integrates a process and product design stage, where optimal process systhesis and experimental data are key to predict the performance of the transfomation stages, and a supply chain design (Martín et al 2023) that requires proper process scale up/down (Sánchez and Martín, 2018) including the concept of economies of numbers, due to the special features of biowaste management, as well as the development of social metrics. We use this methodology to valorize typical biowaste in Spain in ana attemp to create a biuocircualr economy.

Process level is the basic layer of analysis. Food residues are a major challenge. We first focus on the fruit industry. While fruit is itself an added value product, juices and jams are also highly appreciated consumer products. Oranges, apples and grapes are good examples. For instance, an integrated facility for oranges allows producing orange juice, added value food products, fertilizers and utilities. The process begins with juice production. From the peel limonene can be extracted using steam explosion or a solvent, hexane. Next, the waste is digested to produce utilities and fertilizers. The flue gas from the gas turbine can either be used to produce electricity in a regenerative Rankine cycle or as utility within the facility. While the use of steam explosion requires 50% lower production costs, it shows 30% higher investment. The use of hexane consumes 4 times more thermal energy but the processing of the waste requires less digesters. Combined cycles are less profitable. The lower production cost of the steam explosion allows for smaller facilities to become profitable at smaller scales than those using hexane limonene recovery. Grapes produce already a very high value-added product, wine, but the residues can be used to produce a wide range of products, from energy to dyes. The selection of the byproduct production depends on the production capacity that is linked to the wine production where multiproduct refineries are preferred. Apples can be used for juice production or jam and candies elaboration. Linked to the production of apple juice, apple pomace can yield phenolic compounds, pectin and ethanol, that can be used as extraction media for the first one and subcritical water for the second. The sugars released during pectin extraction are used to produce ethanol, although purchasing it is also evaluated. Results show that subcritical water allows easier integration since only energy is required. On-site ethanol production reduces external dependence, showing competitive costs and similar profitability to purchasing at large scales, despite higher electricity consumption. Jam waste is valorized to obtain the same products, but pectin is extracted via citric acid hydrolysis, and an additional stage generates energy via anaerobic digestion (AD) or gasification (GA). Results indicate this hydrolysis requires large amounts of citric acid, which demands substantial sugar inputs, preventing its on-site production. Economically, AD outperforms GA due to lower investment. Finally, renewable-based heating and electrification can reduce the CO2 footprint across both systems but remain less competitive.

The use of lower value bioresidues, can also help reduce our dependence on fossil resources and foreign imports, by producing, pellets, power or synthetic natural has. The methodology is applied in its full extend by first synthesizing the process that valorises biowaste, scale it up/down and formulate a facility location problem for the optimal selection of product, technology and facilities location. The availability of agricultural biowaste make it possible to meet the thermal energy demand of a country like Spain. The problem involves analysing the performance of pellet factories from different residues as well as biomass-based power plants at the process level. Complete demand for thermal energy can be met with an investment of 2175 M€, where up to 47.3% corresponds to palettization and the remaining 52.7% to electricity generation and generating 10 k jobs in rural areas. Going beyond thermal energy, synthetic methane can be produced from lignocellulosic dry residues via gasification and anaerobic digestion of wet waste. The current budget for energy transition would allow achieving a fraction of natural gas self-sufficiency.

Dr. Matjaz Vidmar

Circular Economy and Life-cycle Innovation: From Microalgae to Outer Space

New data-intensive technologies are changing the way we perceive natural environment and shape economic systems. Be it climate change, new materials or natural capital management, systems thinking through data is today at the heart of life-cycle innovation and circural economy entrepreneurship. In this talk, we will examine some of the main trends in Data-Driven Innovation, showcasing experimental projects ranging from using microalgae for decarbonisation, developing new concepts for in-situ resource utilisation (ISRU) in Outer Space, and how Earth Observation is shaping the future of our (trans)planetary ecology. We will also look at tools and strategies that can support a growing circular economy by characterising locally-situated opportunities for responding to global societal challenges.