Integrated Model-Based Frameworks for Synthesis of Anaerobic Treatment Process: Optimizing Operation Using Reactor Networks

Anaerobic treatment technology offers great potential towards achieving Sustainable Development Goals due to its ability to simultaneously breakdown pollutants, generate renewable bioenergy and recycle valuable nutrients from organic waste streams. However, the successful operation of anaerobic digestion (AD) process requires design of optimal process configurations that are well adapted to the characteristics of feedstock available. Due to the highly complex nature of AD, involving multiple reactions with each catalysed by specific groups of microorganisms, multistage anaerobic digestion, in which multiple digesters are operated in a network configuration becomes highly invaluable. This is because such network configurations can optimize overall performance of the AD process by ensuring that the specific conditions under which each reaction step takes place is optimized. In addition, each digester has unique characteristics often making them more adapted to treat waste of specific characteristics than others, and thus utilizing one digester in one configuration may limit the possible combination of pathways, hence limiting overall performance. Model-based design of anaerobic digesters is particularly important as the kinetics captured by AD models can predict operating conditions, volumetric gas production, process stability as well as effluent quality. In addition, systematic model-based approaches for design of anaerobic digestion systems significantly reduce the number of expensive prototype systems and time-consuming studies usually required to obtain and optimal configuration of anaerobic digesters. Even though there exist several studies that use kinetic models to guide design of anaerobic digesters, published literature has primarily been geared towards describing the process of developing a given model to guide design of single stage digester configurations. Remarkably, little research has been carried out on model reliability analysis, especially for the synthesis of multistage digester configurations. This thesis therefore provides both the theoretical background and illustrations (with practical application cases) for development and use of systematic model-based frameworks to guide design and operation of multistage anaerobic digesters irrespective of the information available to the designer. The study uses a methodological approach that develops synergy by systematically coupling model-based techniques (multicriteria decision making tools, practical identifiability, Monte Carlo simulation, adjoint based gradient optimization and attainable region theory) in an optimal framework for synthesis and optimization of anaerobic digester networks. The result of the approach is an optimal framework (decision support system) for synthesis and optimisation of anaerobic digester networks under four practical scenarios: (a) Synthesis based on model requirements or characteristics whereby the study considered the case of no model availability, one-stage kinetic models, two-stage kinetic models, kinetic uncertainty as well as changes in kinetic model structure; (b) Synthesis based on operational/ process objectives, whereby the study considered process stability and process performance (measured in terms of biogas production and organic matter reduction) as design objectives; (c) Synthesis based on economic objectives whereby the study developed digester economic evaluation models based on known economic feasibility indicators as well as macroeconomic parameters; and (d) Synthesis based on feedstock characteristics whereby the study considered two classes of organic substrates (industrial wastewater and animal manure) and analysed the effect of substrate characteristics on the performance targets and optimal configuration of anaerobic digester networks. The results have been captured in five journal publications with the contribution from each paper summarized as follows: Paper 1 presents a framework that uses two-stage kinetic models and process objectives for digester synthesis (mainly methane productivity and volatile solids reduction) while considering the effects of substrate characteristics, using five types of animal manure. The results illustrate that a change in digested substrate significantly influences the operating limits (defined by the attainable region), optimized parameter, as well as the design configuration of the optimal digester structure. This observed substrate effect on attainable regions shows great promises as it paves the way for other substrates such as blackwater, food waste, lignocellulosic waste, as well as co-digested feeds to be considered. Paper 2 presents a framework that uses two-stage kinetic models and stability objectives for digester synthesis (considering inoculum to substrate ratio and instantaneous methanogenic yield) while considering the effects of model structure and sources of inoculum used to startup digester operation. The findings illustrate that the inoculum characteristics influences the structure of the kinetic model used to describe the growth of anaerobic microorganisms and hence the performance targets and digester configurations obtained. Paper 3 presents a framework that uses one-stage kinetic models and economic design objectives for digester network synthesis (developing economic evaluation models based on known economic feasibility indicators as well as macroeconomic parameters), with industrial wastewater as feedstocks. The results illustrate that synthesis of the anaerobic digesters can be tackled using both technical and economic parameters such as payback period as well as country-specific macroeconomic parameters such as interest rate and renewable energy feedin tariff rate. A change in the value of any of these parameters affects the optimal digester configuration. Paper 4 presents a framework that requires no kinetic model for digester synthesis and couples fuzzy multicriteria decision tools with attainable regions for simultaneous synthesis of digester structures and selection of digester subunits considering both techno-economic and environmental aspects. This implies that for the same digester structure, defined in terms of plug flow and continuous stirred tank reactors, the subunits (mainly type of plug flow digester) will differ based on the practical considerations for operating the digester system. Paper 5 presents one of the more significant findings of the study by introducing a framework that simultaneously analyse model reliability, quantifies uncertainty in model states and construct attainable regions that are self-optimizing. Hence, when using attainable regions for performance targeting and digester network synthesis, the results indicate that it is possible to propagate uncertainty of model prediction onto the attainable regions to obtain self-optimizing attainable regions, which is generally smaller than the attainable region but has an advantage of increased robustness. Summarily, the study indicates that using digester networks as opposed to single digesters is able to bypass regions of lower reactivity and improve performance of the anaerobic treatment process. The decision support system should be considered the first point of contact, and used to compliment experiments during planning, design, scale-up and installation of anaerobic digestion plants involving multistage digesters. This will significantly reduce the number of expensive prototype systems and time-consuming studies usually required to obtain an optimal configuration of anaerobic digesters. It is also worth mentioning that even though the study is based on the anaerobic treatment process, the developed frameworks can be applied for synthesis and optimization of other biochemical processes