Designing and performing experiments for model calibration using an automated iterative procedure.
Optimal experimental design for parameter estimation involves complex mathematical and practical steps in order to obtain a model with sufficiently accurate parameters. This paper proposes a methodology where user interaction is only required at the beginning of the experimental design procedure. All subsequent steps are carried out automatically, including: (1) finding the optimal experiment, (2) performing the experiments in practice, and (3) recalibrating the model. A software extension to an existing modelling and simulation package which performs this automatic procedure is also presented. Finally an illustration for the calibration of a one-step nitrification model using respirometric data is given. In this case, the application of the proposed procedure resulted, after three iterations, in considerably better confidence intervals on the parameter estimates, within the desired boundaries.