|Abstract or Summary
- Quantitative procedures were developed to determine the effect of variability in the model parameters required for the estimation of microbial shelflife and thermal processing time. Monte Carlo simulations combined with these predictive models were implemented in Microsoft Excel™. In the first study,
predictive models were used for shelf-life predictions based on the growth of Lactobacillus sakei in meat. The shelf-life values predicted when parameter variability was not considered were 3.6, 115.9, 4.1 and 144.8 h for cases 1 (T = 4°C), 2 (T = 4°C, a[subscript w] = 0.98), 3 (T = 4°C, CO₂ = 2,650 ppm), 4.1 (T = 4°C, a[subscript w] = 0.98,
CO₂ = 2,650 ppm), respectively, whereas 3.9±1.7, 119.4±20.3, 4.6±1.4 and 160.4±0.3 h, respectively, were the values estimated considering parameter variability. The definition of a shelf life with 95% confidence that the product will not fail before the stated expiration date lead to a recommended microbial shelf-life
of 2.5, 100, 3, and 110 h, respectively, When the reported standard deviation of all microbial model parameters describing the effect of the three factors in Case 4 (T = 4°C, a [subscript w] = 0.98, CO₂ = 2,650 ppm) was reduced by 10%, 50% and 90% without
changing mean values, the recommended shelf-life time increased from 110 to 115, 125 and 130 h, respectively. This relatively small increase in the recommended shelf-life, i.e., an increase from 110 to 130 h after a 90% reduction in the variability
of all model parameters, showed that reducing the standard deviation of microbial shelf-life time appeared difficult when assessing the effect of multiple factors.
In a second study, the estimation of a thermal process time at a constant reference temperature (T = 110°C) for the inactivation of Clostridium botulinum spores in commercially produced mushrooms, and based on the reported mean values for thermal inactivation time (D [subscript T]) and initial microbial load (N [subscript o]), yielded a
recommended value of 5.96 min. Unique combinations of generated
N [subscript o]* and D [subscript T]* datasets were used to obtain a distribution of the spore survival probability and the associated percentage of under processing. Next, the coefficient of variation (CV)for the percentage of under processing when using 2 to 500 generated datasets was calculated to determine that 100 was an acceptable minimum number of datasets to estimate 9.6 min as a recommended thermal processing time considering the experimental variability of the parameters D [subscript T] and N [subscript o] and yielding a 10⁻⁹ failure
probability with a 95% confidence. The predictive procedures were used also to assess the impact of reducing the standard deviation (SD) of both No and D110ºC by 10%, 50%, and 90% yielding 8.6, 7.8 and 6.4 min, respectively, as a recommended thermal process time at 110°C with 95% confidence.
In both applications of the procedures here developed, i.e., prediction or microbial shelf-life and design of a thermal process, the user reached a recommendation with 95% confidence using procedures that could be implemented in Microsoft Excel and based on concepts suitable for inclusion in an undergraduate
food science program.