An explanation of non-sterile (leaky) milk packs in well-operated UHT plant

The commercial and widely used single value assessment (SVA) of the risk for the nonsterility of ultra-high temperature (UHT) processed milk is compared with the risk assessment obtained from a Monte Carlo assessment (MCA). The SVA of risk is shown to be of limited real value in accurately predicting the number of non-sterile milk packs. This is because there will not be one single and unique value for each of the UHT process parameters, but rather a distribution of values for each. The effect of distributions in each of the UHT process parameters on a more realistic risk factor P is highlighted through a Monte Carlo calculation with a distribution of practical values attributed to concentration of contaminant spores and thermal resistance, heating temperature, and residence time of the milk in holding section of the sterilizer. Importantly the MCA predictions show that a higher proportion of the number of milk packs will be non-sterile than is predicted by the SVA. This number of nonsterile packs is more or less the number that are anecdotally known to be found non-sterile in well-operated UHT plant, between 1 and 4 in 104. The cause of the non-sterility of these packs is traditionally attributed to leakage of pack material, or seal, and contamination either in processing, or storage and distribution. It is concluded that the occurrence of a fixed number of non-sterile milk operations associated worldwide with UHT plant, and greater than that predicted by the SVA, is the failure to take into account a distribution of values for each of the process parameters. The need for inclusion of a distribution of values for each process parameter is underscored through a comparison of the two methods of calculation of risk. In doing so, it is important to understand, that a risk factor much higher than the simple SVA prediction is highlighted with the MCA. A sensitivity analysis of the MCA prediction of risk could be undertaken to indicate optimum process parameter options for risk reduction in UHT plant. The usefulness of the MCA of risk to similar classes of food, fermentation and pharmaceutical processes should be borne in mind. The approach could be used with more sophisticated microbiological models of thermal inactivation, including non-linearity.