Stone, Karl Michael; (1994) The use of off-line measurement information in on-line state estimation of fed-batch penicillin fermentations. Doctoral thesis (Ph.D), UCL (University College London).

Text The_use_of_off-line_measuremen.pdf Download (24MB)

Abstract

On-line estimation is applied to fermentation processes to provide on-line knowledge of variables for which no reliable sensor is available. These variables usually include the concentrations of the biomass, substrate, and product in the fermentation broth - the variables that are most important for process monitoring and control. As a result of limitations in process understanding, estimator sophistication, and the quality of available measurements, on-line estimation is not always reliable. Measurements of these important variables are, however, usually available from laboratory analyses of samples removed from the fermenter. These are off-line measurements which have the following characteristics; they are irregular and infrequent compared to on-line measurements; there is a delay between sampling the fermenter and the results of the analyses becoming available; and, because these analyses usually involve manual operations, occurrences of inconsistent results are likely. These characteristics often preclude the use of information from off-line measurements in on-line estimation of fermentation processes. In this work the inclusion of information from off-line measurements in the on-line estimation of fermentation processes was investigated. The experimental fermentation system used in the investigation was a small scale, fed-batch fermentation of Penicillium chrysogenum P2. The investigation was performed using two model based, on-line estimators. These represented examples of the simpler and more complex on-line estimators that are commonly applied to fermentation processes. The complex estimator was an implementation of the extended Kalman filter (EKF). Both estimators produced estimates of the biomass concentration which was the variable of greatest interest in this work. Preliminary work was carried out on characterising the measurements important to estimation. Methods for the evaluation of the uncertainties associated with the measurements of the biomass concentration, substrate concentration, the carbon dioxide evolution rate (CER), and the broth volume were devised and their uncertainties evaluated. The identification of inconsistent measurements was also investigated and a consistency test developed. Three methods of including the off-line measurements in both estimators were determined: firstly, reinitialisation of the estimation with the off-line measured values; secondly, the use of off-line measurements in the identification and subsequent adaptation of an influential estimator parameter; and thirdly, a combination of reinitialisation and parameter adaptation. The parameter relating the carbon dioxide evolved as the result of the culture's maintenance activities (mc) adapted in this work. It required off-line measurements of the biomass concentration in its identification. The identification of was limited to the production phase of the fermentation. After an investigation of the three methods it was found that the combination of reinitialisation and parameter adaptation resulted in the best estimation. The use of off-line measurements was found to improve the performance of both estimators compared to their performance without the use of off-line measurements. However, improvement was only possible for conditions of the off-line measurement characteristics that were somewhat idealised compared to those typical of practical fermentation processes: short analysis delays; high off-line measurement frequency, and the elimination of inconsistent off-line measurements. The performance of the estimators deteriorated as the off-line measurement characteristics became less ideal and more typical of practical fermentation processes: long analysis delays; low off-line measurement frequency; and the use of information from inconsistent off-line measurements. The EKF was found to be an unreliable estimator, although it was capable of accurate estimation, it became unstable for no apparent or consistent reason - both with and without the use of off-line measurements.

Download activity - last month

Export as JSONCSVXML

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as CSVJSONXML

Archive Staff Only

View Item