Simulation of enzyme kinetics during malt mashing-A new parameter estimation method

A two-level method for estimation of the parameters of a model describing enzyme kinetics was developed. The model was used to predict the concentrations of three active enzymes during malt mashing. The arising nonlinear optimization problems are difficult to solve and the Levenberg-Marquardt method and Powell's direction set method failed to converge when the initial values for the parameters were not close to the optimum. The two-level method always converged in spite of the chosen initial parameter values and was considerably faster in finding the optimum. The parameter values were estimated for two amylolytic enzymes and one glucanolytic enzyme. The concentrations of active enzymes in wort during laboratory scale mashings with different temperature profiles could be quite accurately predicted by the model.<>