Modeling the Effects of Product Temperature, Product Moisture, and Process Humidity on Thermal Inactivation of Salmonella in Pistachios during Hot-Air Heating

• Published in: Journal of food protection 2021-01, Vol.84 (1), p.47-57
• Main Authors: Casulli, Kaitlyn E, Dolan, Kirk D, Marks, Bradley P
• Format: Article
• Language: English
• Subjects: Colony Count, Microbial; Convection; Convection heating; Deactivation; Dew; Dew point; Flour; Food; Food Handling; Food Microbiology; Food safety; Heating; Hot Temperature; Humidity; Illnesses; Inactivation; Laboratories; Lethality; Model accuracy; Modelling; Moisture content; Moisture effects; Parameter estimation; Pathogens; Pistachio nuts; Pistacia; Process variables; Risk assessment; Root-mean-square errors; Salmonella; Salmonella Enteritidis; Temperature; Variables; Water content
• ISSN: 0362-028X; 1944-9097
• DOI: 10.4315/JFP-20-221

Prior efforts to model bacterial thermal inactivation in and on low-moisture foods generally have been based on isothermal and iso-moisture experiments and have rarely included dynamic product and process variables. Therefore, the objective of this study was to test appropriate secondary models to quantify the effect of product temperature, product moisture, and process humidity on thermal inactivation of Salmonella Enteritidis PT30 on pistachios subjected to dynamic dry- or moist-air heating. In-shell pistachios were inoculated with Salmonella Enteritidis PT30, equilibrated in controlled-humidity chambers (to target water activities [aw] of 0.45 or 0.65), and in some cases, subjected to a presoak treatment prior to heating in a laboratory-scale, moist-air convection oven at multiple combinations (in duplicate) of dry bulb (104.4 or 118.3°C) and dew point (∼23.8, 54.4, or 69.4°C) temperatures, with air speed of ∼1.3 m/s. Salmonella survivors, pistachio moisture content, and aw were quantified at six time points for each condition, targeting cumulative lethality of ∼3 to 5 log. The resulting data were used to estimate parameters for five candidate secondary models that included combinations of product temperature, product moisture, aw, and/or process dew point (coupled with a log-linear primary model). A model describing the D-value as a function of temperature and dew point fit the data well (root mean squared error [RMSE] = 0.86 log CFU/g); however, adding a term to account for dynamic product moisture improved the fit (RMSE = 0.83 log CFU/g). In addition, product moisture content yielded better model outcomes, as compared with aw, particularly in the case of the presoaked pistachios. When validated at the pilot scale, the model was conservative, always underpredicting the experimental log reductions. Both dynamic product moisture and process humidity were critical factors in modeling thermal inactivation of Salmonella in a low-moisture product heated in an air-convection system.