Royce, Patrick N. C.; (1993) Structured monitoring of gas exchange in fermenters for control. Doctoral thesis (Ph.D), UCL (University College London).

Text Structured_monitoring_of_gas_e.pdf Download (15MB)

Abstract

Information from measurements made by on-line sensors are, directly or indirectly, critical to strategies for improved monitoring and control of industrial fermentations. Over the past 20 years, a large body of research has, with little success, attempted to expand the library of on-line measurements routinely used in industrial fermentation. Partly as a result, research efforts have increasingly been targeted at the development of models incorporating on-line data, that describe the dme-profiles of unmeasurable variables of importance to fermentation monitoring. Due to the complexity of most of these models, industrial applications have largely been limited to the simplest of empirical models, based around "derived variables" (that derive directly from one or more on-line measurements), most of them associated with gas exchange, examples being the carbon dioxide evolution rate (CER) and respiratory quotient (RQ). Improvements in the conditioning, analysis and application of gas exchange data would, therefore, be of considerable benefit in improved monitoring, modelling and control of fermentation. This project examines opportunities for such improvements. It was shown that the oxygen transfer rate (OTR) data contain a significant component of uncorrelated Gaussian noise arising from their calculation as a small difference between two large numbers. A chi-square filter was used to frt a linear model to a reduced data set containing only the most recent OTR data, in order to remove this noise. The benefits of applying such a filter were illustrated by the improvement in the quality of OTR data, and related derived variables (the mass transfer coefficient, KLO2a, and the respiratory quotient, RQ), during a Streptomyces clavuligerus fermentation. Theoretical work supported the view that carbon dioxide transfer can be treated as a purely liquid-film limited physical process, as for oxygen. Concerning the error involved in the (widely-used) assumption that the dissolved CO2 partial pressure is equal to the CO2 partial pressure in the exit gas, practical factors were shown to limit the maximum error possible. This error varies with KLO2a, and the aeration rate, being 20-30% in small fermentors, and less in large fermentors. The theoretical results were supported with experimental data from Escherichia coli fermentations. For fermentations run above pH 6.5, the high effective solubility of dissolved carbon dioxide can cause changes in the pH and CER to make unsteady-state terms in the CO2 mass balance important. An effect is to cause the "measured respiratory quotient" as apparent from gas analyses (called here the transfer quotient, or TQ) to differ from the real underlying respiratory quotient (RQ). A model to predict such effects agreed well with experimental results from fermentations of E. coli and S. clavuligerus. The control of pH by on-off additions of acid or base introduces regular fluctuations into the TQ that are not present in the underlying RQ. During exponential growth, the TQ is smaller than the RQ. The RQ can be estimated on-line from the TQ using the model developed. It was shown, both from theory and during an E. coli fermentation, that a simple ratio controller could control the partial pressure of dissolved CO2 to an approximately constant value.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item