Conference Agenda

Biorefinery facilities are a sustainable alternative to fossil fuel production, where valuable biobased materials such as plastics, chemicals and fuels are produced through the conversion of biomass. Green refineries are highly adjustable due to the different combinations of technologies, platforms and substrates available. Appropriate integration of biorefinery systems to industrial facilities would significantly contribute to a more sustainable energy future. However, the adaptability of such facilities complicates the process design often leading to poor decision-making. Even more since biomass feedstocks are inherently diverse through origin, geographic location and storage conditions. Available tools and flowsheeting techniques are often inadequate in establishing the thermodynamic properties of such perplexed organic mixtures. Hence, it’s essential to develop a generalised approach capable of efficiently and accurately profiling non-conventional material and waste streams.

The purpose of this study is to provide standardized models for the chemical characterization of complex streams, ensuring the necessary adaptations while considering the differences in biomass types and forms. This approach provides significant insight in biomass profiling, while allowing for the model to be implemented as a starting point for the design and modelling of biorefinery-associated technologies such as hydrothermal liquefaction. HTL can accommodate a wide range of feedstocks (solid, liquid or even sludge) regardless of their moisture content, thereby eliminating the need for energy intensive pre-treatment making it a viable option for biomass conversion.

This paper conducts a comprehensive analysis of relevant literature to develop an efficient biomass characterization model. Several datasets are gathered and examined to establish a valid representation of the mixture, according to industry accepted standards and laboratory protocols. For reliable property estimation, correlations of key biomass properties are obtained from both computational models and experimental measurements.

A generic mathematical programming approach is followed, using MINLP technologies to efficiently characterize HTL associated material streams (e.g. biomass, biocrude). The problem is formulated with:

Variables that include

• Chemicals in available databases
• Composition in the solution
• Property estimates

Specification parameters that include

• Substrate classification (e.g. breakdown of proteins, sugars, lipids etc.)
• Thermodynamic properties (e.g. densities, HHV, LHV, viscosity etc.)
• Elemental and stoichiometric composition (e.g. ratios of C:H:O:N:S:P)
• Experimental measurements (e.g. moisture content, fixed carbon etc.)

Objective functions featuring

• A vector stream with suitable matching properties and relevance to the nature of the substrate

Integer cuts are implemented to produce classes of solutions of small deviance resulting in alternative populations of components that match the optimization requirements. Integer cuts provide (i) multiple feasible solution points that could establish key chemical components (ii) broader potential for data manipulation (iii) increased flexibility in choosing the appropriate mixture profile.

As sustainability gains global importance, bio-manufacturing pipelines have attracted more attention.^[1] Of particular interest is the valorization of lignocellulosic biomass materials. Studies have demonstrated that Pseudomonas putida can convert lignin into cis,cis-muconic acid, a bioplastics precursor.^[2] This study focuses on the conversion of catechol, a lignin derivative, with the aid of glucose, a growth encouraging substrate, to muconic acid by P.putida.

Cells are extremely complex, with numerous reaction pathways, intermediates, products, and regulatory networks. Precise models of cells are a necessity for optimizing performance and controlling bioprocesses. Current bioprocess phenomenological models, similar to reactor kinetic models, bury information regarding biomass heterogeneity and intercellular reactions within empirical parameters based on biological intuition.^[3] On the other hand, purely machine learning (ML) models have shown to capture the nonlinear complexity of bioprocess but struggle with extrapolation and physical interpretability.^[3] Experimental datasets are often sparse and noisy resulting in poor development and calibration of both these models. To leverage the physical constraints of phenomenological models, and the flexibility and practicality of ML models, hybrid models have been proposed.^[3] In this work, an embedded hybrid modelling structure is explored wherein parameters like growth and consumption rates are modelled using ML models. The hybrid modelling approach will aid in capturing the complex relationships between the external metabolites and the bacteria physiology.

A time variant parameter estimation (TV-PE) technique is employed to train the ML component of the hybrid model. A parameter estimation strategy is explored wherein the errors in the state space and derivative space are minimized to reinforce the learning of the underlying physics behavior. This method is compared to the traditional method of minimizing the error in the state space. Another variation in the hybridization structure is explored: (i) A sequential method where the TV-PE and ML model training is performed separately and (ii) An integrated method where the two steps occur simultaneously. A combination of all these methods is evaluated to determine which can best capture the physics of the bioprocess case study under interpolating and extrapolating scenarios. The hybrid model was shown to consistently outperform purely phenomenological and black box models across various data availability and noise levels.

Moreover, hybrid models lend a framework of interpretability by performing sensitivity analyses on the ML components, providing qualitative insight for biological-intuition guided empirical models for specific phenomenon. Accurate models will aid in process design and optimization and will allow for implementation of more sound and sophisticated control strategies. These are necessary steps for expediting the widespread scale-up and commercialization of bio-manufacturing processes.

References:

[1] J. Wesseler, et al., “Measuring the Bioeconomy: Economics and Policies,” Annu. Rev. Resour. Econ., 2017, pp. 275–298, Oct. 2017.

[2] N.-Z. Xie, et al., “Biotechnological production of muconic acid: current status and future prospects,” Biotechnol. Adv., vol. 32, no. 3, pp. 615–622, May 2014.

[3] A. Tsopanoglou, et al., “Moving towards an era of hybrid modelling: advantages and challenges of coupling mechanistic and data-driven models for upstream pharmaceutical bioprocesses,” Curr. Opin. Chem. Eng., vol. 32, p. 100691, Jun. 2021.

Microalgae cultivation on liquid digestate from the anaerobic co-digestion of agricultural feedstocks is an interesting option for digestate nutrient removal (preventing soil/groundwater pollution) and resource recovery coupled to value-added biomass production. Indeed, some resilient microalgae strains (es. green algae such as Scenedesmus and Chlorella) have been found to show plant bio-stimulating properties, making them particularly suitable for the integration of algal nutrient recovery in a highly growing market.

Altough the process technical feasibility has been already demonstrated in literature at small pilot-scale, no work has been done on the overall process modelling for consequent scenario analysis and techno-economic design optimization, to the best of the authors' knowledge. Indeed, the aim of the authors is also to open novel discussions for plant-wide models’ implementation for this type of biotechnologies.

In this work, a first-principle plant-wide model of the process was developed and thereby described. Two well-established mechanistic, grey-box, physico-chemical and biological three-phase (liquid, solid, gas) models for anaerobic digestion (IWA – ADM1) and algae-based bioremediation processes (ALBA) were considered and modified with necessary equations and extensions to develop a coherent and comprehensive Copp-like interface between the state variables of the two systems, preserving chemical oxygen demand (COD) and atomic mass conservation.

In particular: (i) ADM1 hydrolysis was modified to embrace the co-digestion of agro-zootechnical substrates (es. maize silage, cattle slurry, cattle dung); (ii) inorganic phosphorus dynamic was introduced in ADM1; (iii) main salts' precipitation/dissolution, as well as non-ideality of the multiple acid-base equilibria, were included in both biological models; (iv) the ALBA model was extended with a mechanistic sub-model for simulating the evolution of the raceway pond temperature (also under greenhouse), including the possibility to simulate feedback control closedloop to maintain the culture within strain-specific optimal ranges of temperature.

The resulting system is described by a highly non-linear, stiff DAE system of index-6. So far, it entails also the presence of some non-smooth equations due to the introduction of sign operators in the precipitation/dissolution salt equations (even though those can be easily smoothened). Due to its complexity, the whole plant was implemented in OpenModelica (an open-source software). The implicit and variable-step DASSL DAE solver was exploited as integrator.

Openloop scenario analysis for different upstream co-digester design and operating conditions was carried out to assess the impacts on the downstream microalgae outputs. Yearly dynamic trajectory of reactors' temperature, bacteria concentration/microalgae productivity as well as nutrients removal is reported.

Results highlighted the importance of a proper biorefinery design (with particular care to phosphorous limitation) and yet a noteworthy robustness of the system's performance. The use of the model can facilitate: (i) a more realistic assessment of the technicaleconomic feasibility of the process and (ii) the design of classical and advanced closedloop control strategies.

Futher works involve: (i) the experimental validation of the model for uncertain parameters estimation and (ii) the combination with economic analysis to outline the techno-economic optimization problem of the process design.

To match the growing demand for biomethane production, anaerobic digestors need an optimal management of the input diet. In many cases, co-digestion outcompetes mono-digestion, but is far more complicated to govern, also considering the very limited availability of measurements in full-scale plants, especially for agro-zootechnical plants. The state-of-the-art modelling of the process, the IWA-ADM1, is very useful when it comes to understand and optimize the process, but due to its complexity and overparameterization, literature agrees on its structural and practical unobservability in real-life conditions. Considering the well-known benefits of advanced model-based control, a structurally identifiable reduced-order model called AM2 was derived. Typically, the uncertain parameters of the latter are firstly identified minimizing the simulation error over synthetic ADM1 data, and later refined on real-plant data. Nevertheless, (i) the AM2 model is not suitable to describe the process in the case of coarse/slowly-biodegradable co-substrates and (ii) data synthetically generated or collected without inhibition-active transients cannot be considered fully informative of the non-linearities of the system, so that at least model adaptation by online recursive parameter estimation would be required.

Few works are present in the literature to derive a control-oriented model suitable for agro-zootechnical co-digestion. In addition, no systematic, comprehensive and robust procedures for offline/online uncertain parameter identification are present, to the best of the author’s knowledge.

This work has the novelty to present: (i) the performance comparison between two reduced-order models designed as extensions of the AM2 (hereafter AM2HN and AM2HNtan); (ii) the exploitation of a very informative (yet realistic for full-scale applicability) dataset including transients between different diets and batch activity/biomethane potential tests (initialized with a tailored-made recursive tool); (iii) a parameter subset selection scheme (PSS) based on sensitivity and collinearity analysis used to select only the identifiable parameters for offline identification; (iv) a novel approach to design regularization terms for the online recursive and “moving-horizon”-fashion update of time-varying parameters.

The uncertain parameters of AM2HN and AM2HNtan were firstly estimated using synthetic data generated by an extended ADM1 (hereafter agri-AcoDM) with active acetate and ammonia inhibitions. Only the identifiable parameters pointed out by the PSS were then corrected on real-plant data.

Reasonably, biogas flowrate and composition were the only “continuous” measurements considered. Some spot/”discontinuous” measurement of volatile fatty acids (VFAs), pH and total ammoniacal nitrogen (TAN) was considered. The batch tests were included in the dataset only for the agri-AcoDM parameter estimation, as they can be economically performed realistically few times per year on full-scale inoculum.

Results show narrow 95% confidence intervals from Fisher Information Matrix (FIM) evaluation and interesting fitting performances. As both high-fidelity and reduced-order models fails in fully grasping a VFA accumulation in transient operation, time-varying parameter update was tested and proved to be effective while limiting the risk of overfitting/catastrophic forgetting.

The developed reduced-order models and parameter identification schemes can be exploited in the design/testing of state observers and, eventually, in the design of a nonlinear model predictive control scheme (NMPC) that can be realistically used to optimally control full-scale agricultural anaerobic co-digestion plants.

Despite the considerable potential of biomass as a renewable resource, only a small proportion is currently being converted effectively into bio-based materials. New technologies are essential to build a robust and economically viable bioeconomy. A key target is to valorize available and underutilized raw materials, in particular wastes. In this context, the VALUABLE project [1] aims at demonstrating an innovative platform for the valorisation of Aspergillus Niger biomass deriving from the microbial production of citric acid. This biomass, whose market size is expected to reach 3.3 million tons by 2028, is currently sold as low-value animal feed (300€/t). The focus is the production of multiple value-added products, including yeast oil as a greener alternative to palm oil and non-animal-derived chitosan. Yeast oils will be exploited in cosmetics (i.e., stearates) and alkyd resins (coatings), while non-animal derived chitosan is a multifunctional environmentally and vegan-friendly component for applications in food, agriculture, medicine, pharmaceuticals, and cosmetics.

The process can be divided into six main steps. The approach initially involves an enzymatic solubilisation of the fungal biomass. The resulting sugar-rich aqueous solution is conveyed to an on-site acetic acid production. In the main fermenter, a lipid-rich yeast biomass is grown consuming residual glucose and acetic acid as carbon sources. The produced lipids are released in the following hydrolysis step, and the product is separated into three phases: solid, aqueous phase, and oil phase. On the other hand, the non-solubilized, chitin-rich fraction from the enzymatic solubilization step is deacetylated by a combination of enzymatic and chemical treatment to form chitosan.

Data-driven sub-models for the individual units were developed as plug-ins and integrated in COCO-COFE, a CAPE-OPEN process simulator, to characterize the mass and energy balance of the plant. Adjustable coefficients were tuned to experimental data collected within the project under different operating conditions covering the temperature, residence time, and enzyme content range of interest for commercial applications. A representative chemical formula for each compound, including complex structures of bio-organisms like enzymes and cell mass, was defined based on both in-house data and previous literature to close the mass balance in each conversion step. The process model was exploited to determine the Key Performance Indicators (KPIs) for both productivity and energy consumption, providing a comprehensive overview of the process’ efficiency. The most important finding is that around 180 kg of triglyceride oils and 130 kg of chitosan can be produced from 1 ton of Aspergillus Niger. The total enzyme consumption is 130 kg/ton of fungus. In addition, the production of acetic acid on site through fermentation reduces the need for external acetic acid by 45%.

These estimates illustrate the potential of this innovative approach to producing yeast oil and chitosan. This work establishes the foundational framework necessary for conducting a comprehensive Techno-Economic Analysis (TEA) and Life Cycle Assessment (LCA) of this process, ensuring a thorough evaluation of its economic viability and environmental impact.

[1]: https://valuable-project.eu/

The growing concern over the environmental impacts of the wine industry has driven the search for sustainable technologies to manage its waste, particularly the residual sludge generated during effluent treatment. These sludges, rich in organic matter, represent a significant source of pollution if not properly treated. However, their energy content offers a valuable opportunity to turn this environmental liability into an asset through innovative valorization processes. In this context, hydrothermal liquefaction (HTL) emerges as a promising technology. This process, conducted under subcritical high-temperature and pressure conditions, allows the direct conversion of residual sludge into high-energy-value liquid biofuels. Unlike other treatment methods, HTL can process wet biomass without needing prior drying, making it particularly suitable for managing sludge from wine effluents.

Thus, this research aims to evaluate the conversion of residual sludge, derived from wine effluent treatment, into biofuels through a hydrothermal liquefaction simulation, integrating this process into a sustainable biorefinery for levulinic acid and bioethanol production. The methodology considers the determination of composition and treatment of the wine effluent in order to define a case study is defined. The biorefinery processes as well as the HTL are designed and simulated in Aspen Plus, and also they are evaluated technically and economically. As a result, levulinic acid, sustainable aviation fuel (as product derived from HTL and subsequent bio-oil hydrogenation processes), bioethanol, ethylene glycol, and electrical energy are produced. In addition, the biorefinery reduces the Chemical Oxygen Demand (COD) of the effluent by 99%. In conclusion, valorizing suspended solids from wine effluents through hydrothermal liquefaction is technically and economically feasible. Also, this strategy not only provides an efficient solution for waste management, but also contributes to the transition toward a circular economy by turning waste into energy-rich and value-added products. This research highlights the potential to reduce the wine industry's environmental footprint while generating a renewable energy source.