On-line estimation in a distributed parameter bioreactor: Application to the gluconic acid production

This work presents a methodology which exploits the underlying biochemical structure of bioprocesses to estimate concentrations in aerobic fermenters from oxygen measurements. Although a number of estimators have been proposed over the years in the literature, the methodology proposed in this work is able to operate in transient conditions while does not require the knowledge of the growth kinetics. In addition, it can be also applied to fermenters where the spatial distribution of the concentrations is relevant. In this case, we propose a systematic approach to optimally locate the sensors based on the use of reduced order models. This method allows the reconstruction of the oxygen concentrations from a limited number of sensors. Finally, the methodology proposed will be illustrated on a horizontal tubular reactor for the production of gluconic acid by free-growth of Aspergillus niger. Control of bioreactors has been hampered by a number of obstacles essentially associated with the lack of reliable sensors capable of providing real time measurements of the relevant variables of the process. Additionally, whereas in other fields it is possible to use reliable mathematical descriptions of the processes to design software sensors (observers), this is usually not the case in biotechnology applications where mathematical representations are not well known, specially regarding the reaction rates involved. To overcome these limitations, a theoretical framework which takes advantage of the biochemical reactions of biomass growth and product formation was proposed in (28). The same basic approach was used in (6) for on-line estimation in stirred tank bioreactors. In that work, the authors exploit the underlying reaction structure and transfer mechanisms to systematically design and implement identification schemes for variables and parameters. Other examples for particular types of bioreactors can be seen in the works (12, 11). However, a general description of the methodology together with the precise conditions to apply it in aerobic fermentation still deserves attention. In particular extensions of the methodology to cope with spatial distribution of species concentrations constitute one of those open problems. As discussed in (24) three decades ago, the use of distributed parameter reactors in biotechnology opened new opportunities which gained an increasing interest over the years (32). The mathematical description of such systems relies on the microscopic conservation laws for mass and energy which re- sult into a nonlinear set of partial differential equations (PDE). In most cases the analytical solution is unknown and classical numerical methods for distributed parameter systems (like finite differences or finite element methods) lead to large sets of ordinary differential equations which are computationally involved. This makes the approach unsuitable for real time tasks like on-line estimation, optimization or control (7). In addition, the estimations of the non-measurable variables should be combined with mea- surements of the remaining variables covering the whole spatial domain. This requires having access to a large number of on-line sensors which may be too expensive or physically impossible to be implemented in the desired process (12, 30, 1).