Modeling the dynamics of contamination in the food supply to analyze the impact of intervention strategies

Tejas Himanshu Bhatt, Purdue University

Abstract

The modern food supply is a complex interdependent global system of systems. The increase in complexity results in an inherent inability to analyze the impact of changes to the system. The difficulty of conducting experiments on the real food supply motivated the development of a computational model of the supply chain built using real world data. The three objectives of the study were to: i) develop a computational model of the U.S. food supply, ii) use the model to evaluate the public health and economic impact of contamination scenarios and the intervention strategies used, and iii) analyze the impact of improved traceability and detection and testing methods on the impact of contamination scenarios. A subset of the U.S. food supply was built from bulk ingredient suppliers, food processors, distribution centers, retailers and artificially intelligent consumers. The economic and public health impact of virtual contamination scenarios was then evaluated based on the response action plans deployed by the industry and regulatory agencies. The model was validated by comparing the results of the virtual scenarios to recent food outbreaks. In the worst case scenario, the public health impact of a contaminated ingredient resulted in 91,661 illnesses. The intervention strategies employed by the participants of a simulation exercise, from tracing, holding, testing, releasing or recalling products, resulted in lowering the public health impact to 77,500 illnesses at an economic impact of $638,239,360. The results from the exercises indicate firms closer to the consumers employ a more aggressive intervention strategy during a foodborne crisis. The two types of action plans, conservative and aggressive, captured during simulation events were used to conduct incremental experimental studies on the public health impact of improving traceability and the speed of rapid detection techniques. A reduction in time to test from 4 days to 1 day resulted in the lowering of the public health impact by 52.6%, from 77,500 to 36,759 illnesses. An increase in visibility of traceability data from 1 level forwards and backwards to 4 levels resulted in the lowering of the intensity of the public health impact by 29.5% to 54,607 illnesses. Improving detection allowed a more timely response to the foodborne contamination event, resulting in the earlier peak of the public health impact. Improved traceability afforded a more accurate response to the scenario, resulting in a lower peak in the number of illnesses. The value of using computer simulation models to conduct such large scale analyses is their ability to improve decision-making capabilities by linking actions to their impact in near-real time. For future research work, a more complex supply chain needs to be evaluated in order to validate the findings of this study. Responses to the same scenario should also be recorded with more simulation exercise events in order to make more generalized intervention strategies.

Degree

M.S.

Advisors

Linton, Purdue University.

Subject Area

Food Science

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS