Application of optimal experimental design concept to improve the estimation of model parameters in microbial thermal inactivation kinetics