IDENTIFIKASI PARAMETER MODEL BIOPROSES FED BATCH DENGAN SACCHAROMYCES CEREVISIAE MENGGUNAKAN METODE MAXIMUM LIKELIHOOD

• Main Author: Aisyah, Tita
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/5006
• Institution: Institut Teknologi Bandung

<b>Abstrack</b><p align="justify">Bioprocess is a type chemical process which involving kind of microorganism known as biomass as a catalist that grow in line with the process itself. This process is conducted in a fed-batch operation system, whereby Saccharomyces cerevisiae is used as the biomass, glucose is used as the biomass food. Fermentor is selected as the unit operation. The purpose of this process is to maximize the biomass product and minimize the side effect. This could be achieved by knowing the process dynamics which is formulated in a correct and accurate mathematical model. <br /> The mathematical model of this process is a nonlinear model which is represented in 5 state space equations, 5 output variables and 2 input variables with 4 unknown parameters. The unknown parameters are required to find the accurate mathematical model of this process. Therefore a parameter identification has to be done. The unknown parameter estimatation is conducted by using Maximum Likelihood method. <br /> The Maximum Likelihood is selected because of its effeciency. Estimated parameter structure expressed in a probabilistic as an average value and its covarian matrix so that the valid estimated values will be found. <br /> Sensitivity analysis is conducted to determine the parameter sensitivity rank which will be used as an additional data or information for the identification process. <br /> The identification has been done with the assumption that the process nominal condition is fullfilled. The nominal condition simulation is made to produce input and output data which match with the reference nominal condition. Identification has been done by using 3 varying inputs i.e. Nominal signal, Step signal and 3211 signal. General failure in parameter identification occurs, because of wide parameter's sensitivity differences. This problem could be solved by doing the identification gradually. <br /> Estimated parameter values are found by two step identification with percent standard deviation value (%a ) is less than 13.9 '/o, so that the mathematical model of the fed-batch fermentation process is accurately defined.