Considerations in Calculating Kinetic Parameters From Experimental Data

Abstract

Engineers require quantitative models to design and optimize processes. In the food industry, these process models become very complex because of the unique physical/chemical characteristics and variability of the raw material. Furthermore, frequently data describing rates of reactions and/or changes in foods are generated by food scientists who are not thoroughly familiar with reaction models. Analysis of those data to obtain parameters for reaction models thus becomes critical. In this paper, calculating kinetic parameters from experimental data is examined, and suggestions are presented for determining reaction rates and temperature dependence.

Kinetics of flavour and aroma changes in thermally processed cupua�u (Theobroma grandiflorum) pulp

“…Goodness of ®t criteria (residual plots, R 2 ) were used to decide which model best ®tted each isothermal experiment data, 7 and then the temperature effect was included. The kinetic parameters were determined by a one-step non-linear regression 14,15,7 using Stata Version 3.0 16 statistical software. This procedure narrowed the con®dence intervals of the parameters estimated due to the increased degrees of freedom.…”

Stability of cupuaçu (Theobroma grandiflorum) nectar during storage

“…A one-step procedure was next tried to reduce the standard deviation in k ref and E a , by performing a non-linear regression through all the data points in order to calculate E a and k ref from the original data (Arabshahi & Lund, 1985). The Statistical Software (Stata Corporation, 1995) was used.…”

Kinetics of Food Deterioration and Shelf-Life Prediction

“…Arabshahi and Lund (1985) proposed appropriate regression weight factors that can be used in this case.…”

“…If a linear regression method is used to estimate the parameters, their 95% confidence limits can be calculated using the Student t distribution. In addition to the confidence limits, a list of standarized residuals and a residual plot is a useful statistical tool that allows evaluation of how well the chosen equation can model the data and also permits the recognition of extreme or outlier values that may be the result of experimental errors or other extraneous effects and should be excluded from the calcualtions (Arabshasi and Lund, 1985). The standarized residuals should be randomly distributed around zero and usually within -2 and +2.…”