Final Report Summary - DOP-ECOS (Optimal Design and Operation of Microbial Ecosystems for Bioenergy Production and Waste Treatment)
Many microbial ecosystems, as part of their normal routine, have the potential to provide services to society and improve environmental quality. Some can degrade contaminants that pollute water, air or soil. Others can transform waste materials into valuable renewable resources, including bioenergy, biomaterials and high-value products. This generic capability opens the possibility for combining several microbial ecosystems into integrated bioprocesses or biorefineries, where various types of bioenergy or biomaterials are produced and multiple sources of pollution are treated, all at the same time.
The focus in DOP-ECOS has been more specifically on bioprocesses that couple a photobioreactor, where microalgae capture sunlight to produce new biomass and lipids, and an anaerobic digester, where bacteria convert biomass into biogas and recover nutrients (Figure 1). Its general objective were two-fold: (i) to optimize the design, operation and control of integrated microalgal/bacterial processes; and (ii) to develop the supporting methods and tools for their reliable analysis and optimization. With experimental research and demonstration programs are carried out worldwide to identify suitable algae strains and expand algal biofuel production from a craft to a major industrial process, a distinctive feature of DOP-ECOS has been to exploit this experimental information through the development of reliable mathematical models that can be used with systematic optimization methods in order to assess the potential and realize the most out of these processes.
The first work-package of DOP-ECOS was concerned with the development of new algorithms for the efficient and reliable estimation, optimization and control of biotechnological processes. These problems are particularly challenging in that the mathematical models at hand usually contain a large number of parameters, which may carry a significant amount of uncertainty. Our main focus throughout DOP-ECOS has been on complete search methods that can provide a certificate that the best possible solutions are computed, as opposed to local, possibly suboptimal, solutions. A key contribution in this area has been the development and analysis of improved techniques for bounding the set of all trajectories of a dynamic system under parametric uncertainty, a notoriously difficult problem. The applications of this generic bounding capability into global and robust optimization algorithms was also investigated as part of this work-package. Particular emphasis has been on set-membership estimation, a technique for determining the set of all possible parameter values of a model in order for its predictions to be consistent with a set of scarce/noisy measurements – as opposed to determining a single parameter estimate that gives a good fit. Another key contribution has been a new algorithmic framework, called branch-and-lift, that enables guaranteed global optimization of optimal control problems, without the need for a priori discretization of their state or control variables. All the computer programs/libraries developed through DOP-ECOS are in the form of a C++ library called CRONOS (Complete seaRch sOlutions for NOnlinear Systems), which is publicly available from the repository: provisional release date by the end of 2015. Overall, these algorithmic developments are particularly significant in that they open the perspective of applying global and robust optimization technology to industrially relevant problems, such as biotechnological applications.
The second work-package was concerned with the application of these advanced optimization-based methods and tools for the robust design and control of integrated microalgal-bacterial processes. Optimizing the performance of the anaerobic digester and of the photobioreactor separately, using state-of-the-art models, revealed that the lack of accurate mathematical models, especially for the photobioreactor, was an important limitation for such approaches. Our focus therefore has been on developing mathematical models capable of quantitative prediction of microalgae growth in large-scale production systems. A first important contribution here has been the development of a mathematical model capable of quantitative prediction of the state of the photosynthetic apparatus of microalgae under realistic light conditions (Figure 2). Among the possible applications, this model can be used to predict PI-response curves, a popular way of inferring microalgae culture productivity, based solely on fast fluorescence measurements; another application of the model is for guiding the selection or use of genetically engineering algae strains. A second key contribution in this work-package has been the development of multiscale models integrating the fast photosynthetic processes with slower processes of acclimation and cell growth (time-scale of days). The process of photoacclimation, which refers to the adjustment of the light harvesting capacity of the microalgae to the incoming irradiance, is indeed particularly important in outdoor culture systems. Our final contribution has been the integration of the developed microalgae growth models into multiphysics models for the prediction of large-scale productivities in raceway ponds. Scaling up microalgae production systems is particularly challenging due to the presence of light- and nutrient-dependent processes that are competing for growth. In contrast to lab-scale experiments, full-scale production systems can exhibit a dramatic loss of productivity due to imperfect mixing or light distribution, contamination or lack of adequate monitoring and control. Our focus has been on quantifying the effects of imperfect light distribution on growth. More specifically, we successfully combined the microlagae growth model with light-attenuation models describing the vertical light distribution in a raceway pond and computational fluid dynamics (CFD) models describing the flow conditions (Figure 3). This generic simulation capability is particularly significant as it allows identifying the main bottlenecks and improvement potential in large-scale culture systems. Moreover, it provides an ideal ground for testing hypotheses regarding the use of genetically modified microalgae species, or for determining the best location for a microalgae farm based on geographical information systems (GIS).
For more information, please visit the fellow's website at: http://www3.imperial.ac.uk/environmentenergyoptimisation
The focus in DOP-ECOS has been more specifically on bioprocesses that couple a photobioreactor, where microalgae capture sunlight to produce new biomass and lipids, and an anaerobic digester, where bacteria convert biomass into biogas and recover nutrients (Figure 1). Its general objective were two-fold: (i) to optimize the design, operation and control of integrated microalgal/bacterial processes; and (ii) to develop the supporting methods and tools for their reliable analysis and optimization. With experimental research and demonstration programs are carried out worldwide to identify suitable algae strains and expand algal biofuel production from a craft to a major industrial process, a distinctive feature of DOP-ECOS has been to exploit this experimental information through the development of reliable mathematical models that can be used with systematic optimization methods in order to assess the potential and realize the most out of these processes.
The first work-package of DOP-ECOS was concerned with the development of new algorithms for the efficient and reliable estimation, optimization and control of biotechnological processes. These problems are particularly challenging in that the mathematical models at hand usually contain a large number of parameters, which may carry a significant amount of uncertainty. Our main focus throughout DOP-ECOS has been on complete search methods that can provide a certificate that the best possible solutions are computed, as opposed to local, possibly suboptimal, solutions. A key contribution in this area has been the development and analysis of improved techniques for bounding the set of all trajectories of a dynamic system under parametric uncertainty, a notoriously difficult problem. The applications of this generic bounding capability into global and robust optimization algorithms was also investigated as part of this work-package. Particular emphasis has been on set-membership estimation, a technique for determining the set of all possible parameter values of a model in order for its predictions to be consistent with a set of scarce/noisy measurements – as opposed to determining a single parameter estimate that gives a good fit. Another key contribution has been a new algorithmic framework, called branch-and-lift, that enables guaranteed global optimization of optimal control problems, without the need for a priori discretization of their state or control variables. All the computer programs/libraries developed through DOP-ECOS are in the form of a C++ library called CRONOS (Complete seaRch sOlutions for NOnlinear Systems), which is publicly available from the repository: provisional release date by the end of 2015. Overall, these algorithmic developments are particularly significant in that they open the perspective of applying global and robust optimization technology to industrially relevant problems, such as biotechnological applications.
The second work-package was concerned with the application of these advanced optimization-based methods and tools for the robust design and control of integrated microalgal-bacterial processes. Optimizing the performance of the anaerobic digester and of the photobioreactor separately, using state-of-the-art models, revealed that the lack of accurate mathematical models, especially for the photobioreactor, was an important limitation for such approaches. Our focus therefore has been on developing mathematical models capable of quantitative prediction of microalgae growth in large-scale production systems. A first important contribution here has been the development of a mathematical model capable of quantitative prediction of the state of the photosynthetic apparatus of microalgae under realistic light conditions (Figure 2). Among the possible applications, this model can be used to predict PI-response curves, a popular way of inferring microalgae culture productivity, based solely on fast fluorescence measurements; another application of the model is for guiding the selection or use of genetically engineering algae strains. A second key contribution in this work-package has been the development of multiscale models integrating the fast photosynthetic processes with slower processes of acclimation and cell growth (time-scale of days). The process of photoacclimation, which refers to the adjustment of the light harvesting capacity of the microalgae to the incoming irradiance, is indeed particularly important in outdoor culture systems. Our final contribution has been the integration of the developed microalgae growth models into multiphysics models for the prediction of large-scale productivities in raceway ponds. Scaling up microalgae production systems is particularly challenging due to the presence of light- and nutrient-dependent processes that are competing for growth. In contrast to lab-scale experiments, full-scale production systems can exhibit a dramatic loss of productivity due to imperfect mixing or light distribution, contamination or lack of adequate monitoring and control. Our focus has been on quantifying the effects of imperfect light distribution on growth. More specifically, we successfully combined the microlagae growth model with light-attenuation models describing the vertical light distribution in a raceway pond and computational fluid dynamics (CFD) models describing the flow conditions (Figure 3). This generic simulation capability is particularly significant as it allows identifying the main bottlenecks and improvement potential in large-scale culture systems. Moreover, it provides an ideal ground for testing hypotheses regarding the use of genetically modified microalgae species, or for determining the best location for a microalgae farm based on geographical information systems (GIS).
For more information, please visit the fellow's website at: http://www3.imperial.ac.uk/environmentenergyoptimisation
