Scalability of a Discrete Element Model for Salmonella Cross-contamination in Granular Low-water Activity Foods

Introduction: Modeling cross-contamination of bacteria in granular low-moisture foods is a particular challenge due to the discrete nature of these materials. Scaling-up models based on laboratory data to industrial-scale system is limited because of the lack of first-principle models. An ideal cross-contamination model should enhance its scalability, so that it can be utilized for an industrial-scale system without further validation burden.

Purpose: The purpose of this study was to assess the scalability of a discrete element method (DEM) model of bacterial cross-contamination to industrial-scale systems.

Methods: Almond kernels were inoculated with Salmonella Enteritidis PT30 and mixed with clean almonds in a rotating drum at a bench top scale of ~200 g (5 g of inoculated almonds). A DEM bacterial transfer model was developed from these results and validated against a pilot-scale model of ~1 kg. After validation, the model was used to simulate an industrial-scale scenario of ~200 kg of almonds mixed with 5 kg of contaminated almonds.

Results: The lab-scale experiments (with contaminated almonds at ~8.3 log CFU/g) yielded 4.3±0.2 log (CFU/g) maximum transferrable bacterial load after 600 s at 8 rpm. The calibration model of the experiment was fit to the data (RMSE=0.005 log CFU/g) and validated with pilot-scale data sets (RMSE=0.057 log CFU/g). The results for the 200 kg rotary batch mixer simulated a similar trendline as actual experiments, showing a maximum transferrable bacterial load of 4.1±0.1 log (CFU/g) after 600 s at 8 rpm, and demonstrated reasonable scalability of the DEM model.

Significance: DEM modeling appears to be an efficient tool to model the interactions of particulate low-moisture food products. The scalability of the DEM model will contribute to risk modeling associated with bacterial cross-contamination scenarios.