This research is aimed at assessing the quantitative risks involved with an ethanol pipeline. Pipelines that run from the Midwest, where the vast majority of ethanol is produced, to the target areas where reformulated gasoline is required (California, Texas Gulf Coast, New England Atlantic Coast) will be of particular interest. The goal is to conduct a quantitative risk assessment on the pipeline, truck, and rail transportation modes to these areas. As a result of the quantitative risk assessment, we are able to compare the risk associated with the different modes of transportation for ethanol. In order to perform and compare the quantitative risk assessment, the following challenges are addressed: • Identify target areas requiring reformulated gasoline • Map detailed route for each transportation mode to all three target areas • Perform a quantitative risk assessment for each transportation mode • Compare quantitative risk assessment results for each route and transportation mode The focus is on California, Texas Gulf Coast, and New England Atlantic Coast because of the large volume. It is beneficial to look at these areas as opposed to the smaller areas because pipeline transportation requires very large volumes. In order to find a meaningful comparison between all three transportation modes, only the areas with the three large volumes were evaluated. Since the risk assessment is completed using historical data, each route is segmented in a way that is consistent with the data that is available. All of the curves support the hypothesis that pipeline transportation poses the least societal risk when transporting ethanol from the Midwest to target areas. Rail transportation poses the largest amount of societal risk. While overall rail incidents are not as frequent as road incidents, the frequency of a fatality is much higher when an incident does occur.

This research is aimed at assessing the quantitative risks involved with an ethanol pipeline. Pipelines that run from the Midwest, where the vast majority of ethanol is produced, to the target areas where reformulated gasoline is required (California, Texas Gulf Coast, New England Atlantic Coast) will be of particular interest. The goal is to conduct a quantitative risk assessment on the pipeline, truck, and rail transportation modes to these areas. As a result of the quantitative risk assessment, we are able to compare the risk associated with the different modes of transportation for ethanol. In order to perform and compare the quantitative risk assessment, the following challenges are addressed: 1) Identify target areas requiring reformulated gasoline 2) Map detailed route for each transportation mode to all three target areas 3) Perform a quantitative risk assessment for each transportation mode 4) Compare quantitative risk assessment results for each route and transportation mode The focus is on California, Texas Gulf Coast, and New England Atlantic Coast because of the large volume. It is beneficial to look at these areas as opposed to the smaller areas because pipeline transportation requires very large volumes. In order to find a meaningful comparison between all three transportation modes, only the areas with the three large volumes were evaluated. Since the risk assessment is completed using historical data, each route is segmented in a way that is consistent with the data that is available. All of the curves support the hypothesis that pipeline transportation poses the least societal risk when transporting ethanol from the Midwest to target areas. Rail transportation poses the largest amount of societal risk. While overall rail incidents are not as frequent as road incidents, the frequency of a fatality is much higher when an incident does occur.

Dust and hybrid mixture explosions continue to occur in industrial processes that handle fine powders and flammable gases. Considerable research is therefore conducted throughout the world with the objective of both preventing the occurrence and mitigating the consequences of such events. In the current work, research has been undertaken to help advance the field of dust explosion prevention and mitigation from an emphasis on hazards to a focus on risk. Employing the principles of quantitative risk assessment (QRA) of dust and hybrid mixture explosions, a methodological framework for the management of these risks has been developed. The Quantitative Risk Management Framework (QRMF) is based on hazard identification via credible accident scenarios for dust explosions, followed by probabilistic Fault Tree Analysis (using Relex – Reliability Excellence – software) and consequence severity analysis, represented by maximum explosion pressure, (using DESC – Dust Explosion Simulation Code – software). Identification of risk reduction measures in the framework is accomplished in a hierarchical manner by considering inherent safety measures, passive and active engineered devices, and procedural measures. Dust explosion tests to determine icing and granulated sugar dust explosibility characteristics have been achieved in a 20-L Siwek chamber, and, accordingly, DESC fuel files were built and DESC has validated. Three industrial case studies are presented to show how the QRMF could has been helpful in reducing dust and hybrid mixture explosion risk at the Imperial Sugar refinery, the Semabla grain storage silo, and a hypothetical 400-m3 polyethylene storage silo. DESC simulations and Probit equation for structural damage provide the destructive percentage of each pressure zone in the simulations, followed by probabilistic FTA that were achieved for the first two case studies, before and after applying the framework. Detailed individual and societal risks calculations were made and F-N curves plotted for the two processes. The polyethylene silo case study is presented to show how inherent safety measures can be helpful in reducing dust and hybrid mixture explosion risk. The framework showed significant risk reduction to the point where the residual risks are acceptable for both processes. Finally, assessment results are compared and improvements measured. / The research main objective is: to manage the risks of any expected dust or hybrid mixture explosion in industrial complexes (large-scale).

Made available in DSpace on 2014-06-12T17:42:48Z (GMT). No. of bitstreams: 2 arquivo7684_1.pdf: 3197582 bytes, checksum: e9a5a1df323907fb05bb3c1b206c2fd5 (MD5) license.txt: 1748 bytes, checksum: 8a4605be74aa9ea9d79846c1fba20a33 (MD5) Previous issue date: 2011 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Recentes acidentes industriais, como vazamentos tóxicos, têm causado danos catastróficos ao meio ecológico (i.e. plantas e animais), de modo que um método efetivo para analisar riscos ecológicos tem sido demandado. Em primeiro lugar, este trabalho tem como objetivo propor uma metodologia capaz de quantificar riscos ecológicos inerentes a eventos raros como acidentes industriais. Utiliza-se a modelagem populacional para simular futuras mudanças na abundância populacional de espécies-chave em risco e, assim, estimar a probabilidade de extinção ou declínio, tempo para extinção e outras medidas, para cada cenário acidental. Assim, foi possível desenvolver uma abordagem que combina os danos ecológicos (previstos através da modelagem populacional) com a frequência de ocorrência do cenário acidental (estimada através de dados históricos e análise de confiabilidade). O resultado é uma curva de risco FN (similar ao resultado de uma análise de risco a humanos), onde N é o declínio populacional médio e F a frequência acumulada de acidentes com declínio maior ou igual a N. Em segundo lugar, o trabalho apresenta uma aplicação da metodologia para quantificar os riscos ecológicos provenientes de acidentes associados ao transporte e manuseio de petróleo que abastece uma refinaria no Complexo Industrial Portuário de Suape-PE, no Nordeste do Brasil. Esta instalação está localizada próxima a um rico ecossistema aquático de alta biodiversidade. A população de uma espécie nativa foi estrategicamente escolhida para representar o ecossistema, alguns cenários de derramamento de petróleo foram simulados e suas frequências de ocorrência estimadas. Para cada cenário acidental, a concentração de óleo que atinge a população foi prevista via modelagem de destino e transporte. Os riscos ecológicos foram quantificados e apresentados em uma curva FN. Uma análise de sensibilidade foi feita para explorar como mudanças em parâmetros específicos causam mudanças nas medidas de risco. Além disso, a incerteza foi medida como um intervalo (limite superior e inferior) para as medidas de riscos com base em cenários pessimistas e otimistas. Finalmente, a metodologia mostrou-se viável, eficiente, conveniente e flexível, apesar de que algumas melhorias ainda podem ser feitas e estas foram propostas para trabalhos futuros

Hazardous Materials (HazMat) are transported throughout Canada in a great number of road shipments. The transportation of HazMat poses special risks for neighboring population and environment. While HazMat accidents are rare events, they could be catastrophic in nature and could result in substantial damage to nearby communities. Effective emergency response plays an important role in the safe transportation of HazMat. Transportation of HazMat involves different parties, including shippers, regulators, and surrounding communities. While the shipping party is responsible for safe delivery of HazMat shipments, it is the responsibility of local emergency service agencies to respond to accidents occurring within their jurisdictions. In this research, the emergency response to HazMat transport accidents is assumed to be delegated exclusively to specially trained and equipped HazMat teams. This research proposes a new comprehensive systematic approach to determine the best location of HazMat teams on regional bases utilizing HazMat transport risk as a location criterion. The proposed model is the first to consider emergency response roles in HazMat transport risk analysis, and was intended as an optimization tool to be used by practitioners for HazMat emergency response planning. Additionally, the proposed model can be used to assess risk implications in regards to current locations of HazMat teams in a region, and to develop effective strategies for locating HazMat teams, such as closing and/or relocating teams in the region. The model investigates how HazMat team locations can be tailored to recognize the risk of transporting HazMat and would provide a more objective set of input alternatives into the multi-criteria decision making process of regionally locating HazMat teams. The proposed model was applied to the region of southwestern Ontario in effort to illustrate its features and capabilities in the HazMat emergency response planning and decision making process. Accordingly, the model provided very useful insights while reviewing several HazMat team location strategies for the southwestern Ontario region and investigating tradeoff among different factors. This research contributes to a better understanding of emergency response roles by reducing HazMat transport risks, and will greatly benefit both researchers and practitioners in the field of HazMat transport and emergency response.

Manitoba’s flooding is one of the principal sources of costs associated with flood disasters. Despite the tremendous financial investments in structural mitigation, flood cost, continue to rise. Currently at the national and provincial level, there exists no standardized operational framework and tool for flood risk assessment. The primary role of this research was two-fold. First, Hazus-MH a Quantitative Risk Assessment (QRA) tool was evaluated as a potential flood-loss estimation model in a Manitoba context. Secondly, to examine the application of a QRA tool in policy, a face-to-face questionnaire was completed with flood experts in municipal and provincial government and with consultants. The study results suggested that Hazus-MH has the potential to be applied as a standard QRA tool in Manitoba. This research found that developing a standard QRA tool in Manitoba, would empower communication between decision-makers, centralize data to support disaster planning, and reduce the cost of recovery. / May 2016

Hazardous Materials (HazMat) are transported throughout Canada in a great number of road shipments. The transportation of HazMat poses special risks for neighboring population and environment. While HazMat accidents are rare events, they could be catastrophic in nature and could result in substantial damage to nearby communities. Effective emergency response plays an important role in the safe transportation of HazMat. Transportation of HazMat involves different parties, including shippers, regulators, and surrounding communities. While the shipping party is responsible for safe delivery of HazMat shipments, it is the responsibility of local emergency service agencies to respond to accidents occurring within their jurisdictions. In this research, the emergency response to HazMat transport accidents is assumed to be delegated exclusively to specially trained and equipped HazMat teams. This research proposes a new comprehensive systematic approach to determine the best location of HazMat teams on regional bases utilizing HazMat transport risk as a location criterion. The proposed model is the first to consider emergency response roles in HazMat transport risk analysis, and was intended as an optimization tool to be used by practitioners for HazMat emergency response planning. Additionally, the proposed model can be used to assess risk implications in regards to current locations of HazMat teams in a region, and to develop effective strategies for locating HazMat teams, such as closing and/or relocating teams in the region. The model investigates how HazMat team locations can be tailored to recognize the risk of transporting HazMat and would provide a more objective set of input alternatives into the multi-criteria decision making process of regionally locating HazMat teams. The proposed model was applied to the region of southwestern Ontario in effort to illustrate its features and capabilities in the HazMat emergency response planning and decision making process. Accordingly, the model provided very useful insights while reviewing several HazMat team location strategies for the southwestern Ontario region and investigating tradeoff among different factors. This research contributes to a better understanding of emergency response roles by reducing HazMat transport risks, and will greatly benefit both researchers and practitioners in the field of HazMat transport and emergency response.

Orientadores: Pilar Rodriguez de Massaguer, Amauri Rosenthal / Dissertação (mestrado) - Universidade Estadual de Campinas, Faculdade de Engenharia de Alimentos / Made available in DSpace on 2018-08-08T21:56:14Z (GMT). No. of bitstreams: 1 Sant'Ana_AndersondeSouza_M.pdf: 2935543 bytes, checksum: 8836bcdf5bfa2652ee4e277718b808dc (MD5) Previous issue date: 2007 / Resumo: A patulina é uma micotoxina produzida por fungos pertencentes aos gêneros Penicillium, Aspergillus e Byssochlamys. Penicillium expansum se destaca por ser potencial produtor de patulina nas maçãs, enquanto Byssochlamys nivea e B.fulva se destacam pela potencial produção desta micotoxina em sucos de maçãs pasteurizados. À patulina tem sido atribuídos diversos efeitos agudos e crônicos adversos à saúde humana. No presente estudo determinou-se quantitativamente o risco dos níveis de patulina no suco de maçã ultrapassar o limite de 50ppb estabelecido pela Organização Mundial de Saúde (OMS) e a probabilidade de produção da patulina por bolores termoresistentes sobreviventes à pasteurização do produto. Para isto, i) avaliou-se a ocorrência de bolores termoresistentes e patulina em amostras pertencentes à 5 diferentes lotes de suco de maçã provenientes de uma unidade produtora localizada no sudeste do Brasil; ii) avaliou-se a capacidade de produção da patulina por cepas de B.fulva (IOC 4518) e B.nivea (ATCC 24008 e FRR 4421) em suco de maçã armazenado à 21ºC e à 30ºC (consideradas temperaturas médias anuais das regiões tropicais e subtropicais do Brasil); iii) determinou-se a cepa Byssochlamys produtora de patulina mais termoresistente dentre as três estudadas, em suco de maçã; iv) determinou-se a resistência térmica através do método dos tubos TDT (thermal death tubes) da cepa mais termoresistente; v) Estabeleceu-se o efeito da pasteurização em sistema contínuo (UHT), simulando as condições industriais, sobre a cepa de Byssochlamys spp mais termoresistente produtora de patulina; vi) determinou-se a probabilidade de produção de patulina pela cepa mais termoresistente produtora de patulina em suco clarificado de maçã variando-se a temperatura de estocagem (21ºC e 30ºC), com carga remanescente pós-processo de 10º e 101 esporos / 100 mL e vii) determinou-se quantitativamente o risco da patulina em suco clarificado de maçã a partir de 15 cenários que relacionados a diferentes cargas de esporos sobreviventes e temperatura de estocagem do suco pós-pasteurização, utilizando-se a modelagem de Monte Carlo, feita através do software @RISK versão 4.5 for students com 1 simulação e 10000 iterações. Os resultados revelaram que a ocorrência de bolores termoresistentes nas amostras de suco de maçã é baixa (<10ºesporos/100mL), com a cepa isolada de (Aspergillus carneus ¿ IOC 4519) não sendo confirmada como um bolor termoresistente. As três cepas estudadas (B.nivea FRR 4421, B.nivea ATCC 24008 e B.fulva IOC 4518) foram capazes de produzir a patulina em concentrações que dependeram da carga de esporos inoculada no suco de maçã, temperatura de estocagem e material de embalagem. B.fulva IOC 4518 foi determinada como a cepa mais termoresistente, sobrevivendo ao choque térmico de 95ºC por 5 min. Valores D* à 85ºC, 90ºC, 92ºC e 95ºC iguais a 64,58 min; 16,68 min; 6,31 min e 3,10 min, respectivamente foram obtidos, enquanto o valor Z foi igual a 7,4ºC. O processo de pasteurização do suco de maçã na unidade Microthermics mostrou que há variabilidade com relação ao número de reduções causadas pelo processo equivalente ao industrial quando variações na temperatura de processo da ordem de até 1ºC são observadas. Maiores probabilidades de crescimento de B.fulva IOC 4518 e maior extensão da deterioração no suco de maçã estão relacionadas ao aumento da carga de esporos sobreviventes e temperatura de estocagem pós-processo. A produção de patulina por B.fulva IOC 4518 é influenciada principalmente pela temperatura de estocagem quando a carga de sobreviventes é elevada (101esporos/100mL), com maiores quantidades da micotoxina sendo produzidas à 30ºC do que à 21ºC. Através do modelo de risco para o suco de maçã e patulina pôde-se concluir que a etapa de recepção é sempre a que mais impactou para que níveis elevados de patulina estejam presentes nos sucos de maçã. Mas, a etapa de estocagem após a pasteurização quando há bolores termoresistentes sobreviventes, foi a responsável pelas maiores concentrações finais de patulina, à medida que se aumentava o tempo de estocagem. As etapas de lavagem, filtração e seleção das frutas, respectivamente, são as responsáveis pela redução nos níveis de patulina durante o processamento do suco, enquanto a pasteurização em virtude da elevada resistência térmica desta micotoxina praticamente não apresenta efeito algum para redução dos níveis da micotoxina no produto final. Após 10000 iterações, dentre os 15 cenários avaliados, somente nos de número 1 (sem sobreviventes à pasteurização e concentração de patulina na matéria-prima na faixa baixa), 2 (sem sobreviventes à pasteurização e concentração de patulina na matéria-prima na faixa média), 4 (10ºesporos/100mL de B.fulva IOC 4518 sobreviventes à pasteurização, com temperatura de estocagem de 21ºC e concentração de patulina na matéria-prima na faixa baixa) e 7 (10ºesporos/100mL de B.fulva IOC 4518 sobreviventes à pasteurização, com temperatura de estocagem de 30ºC e concentração de patulina na matéria-prima na faixa baixa) o limite de 50ppb de patulina não seria ultrapassado para nenhuma das iterações, considerando-se valores médios e máximos da concentração final desta micotoxina. A distribuição da concentração de patulina no produto final se ajustou às distribuições do tipo Betageneral, Lognormal e Inversa Gaussiana dependendo da concentração de patulina na recepção das frutas, com a média e a maior parte dos dados se concentrando à esquerda / Abstract: Patulin is a mycotoxin produced by some species from Penicillium, Aspergillus and Byssochlamys ssp genera. Penicillium expansum are known by their potential to produce patulin in apples, while Byssochlamys nivea and B.fulva are recognized by its potential to produce this micotoxin in pasteurized apple juice. Several acute and chronic effects to human health have been attributed to patulin. This study has quantitatively assessed the risk of levels of patulin to exceed the level established by The World Health Organization ¿ WHO (50ppb) and also was determined the probability of patulin being produced by heat-resistant mold which survived the apple juice pasteurization. Therefore, the following items have been analyzed: i) it was evaluated the occurrence of heat resistant mold and patulin in samples belonging to 5 different lots of apple juice from a factory located in the southeast of Brazil; ii) the ability of patulin production by the strains of B.fulva (IOC 4518) and B.nivea (ATCC 24008 and FRR 4421) in apple juice stored at 21°C and 30°C (these are average year temperature in the tropical and subtropical regions of Brazil); iii) it was determined which of the three patulin producer strains, B. fulva and/or B.nivea, was the most heat resistant in apple juice; iv) it was determined the heat resistance (D and Z values) of the most heat resistant patulin producer strain of Byssochlamys using thermal death tubes (TDT); v) the effect of the continuous pasteurization system (UHT), simulating the industrial conditions, has been established over the most heat resistant and patulin producer strain of Byssochlamys spp; vi) the probability of producing patulin by the most heat resistant strain in clarified apple juice stored at 21°C and 30°C with a survival level post pasteurization process of 10º e 101 spores/100 mL, has also been established. vii) it has been quantitatively assessed the risk of patulin in clarified apple juice using Monte Carlo simulation, with @Risk software for students (version 4.5). The simulation was carried out with 10000 iterations. The results showed that the occurrence of heat resistant mold in the apple juice samples examined was low (<10ºesporos/100mL), with the strain Aspergillus carneus ¿ IOC 4519 isolated not confirming their heat resistance. The three mold strains studied (B.nivea FRR 4421, B.nivea ATCC 24008 and B.fulva IOC 4518) were able to produce patulin in concentrations that were dependent of spore inocula in apple juice, storage temperature and package type. B.fulva IOC 4518 was determined as the most heat resistant strain, surviving to heat shock at 95ºC/5 min. D* values at 85ºC, 90ºC, 92ºC and 95ºC of 64,58 min; 16,68 min; 6,31 min and 3,10 min, respectively were obtained, while z value was of 7.4ºC. The apple juice pasteurization process applied in a Microthermics pilot plant showed variability related to the number of decimal reductions caused by the equivalent process when temperature variations were near 1ºC. Higher growth probabilities for B.fulva IOC 4518 and higher extension of spoilage of apple juice are related to the increase of survival spore level and to storage temperature pos-pasteurization. The patulin production by B.fulva IOC 4518 was mainly influenced by storage temperature when survival spores level is elevated (101spores/100mL), with the higher quantities of this mycotoxin being produced at 30ºC than 21ºC. The risk assessment model for the apple juice and patulin showed that fruit reception is always the step that more impacted to higher levels of patulin being found in apple juices. However, storage step after pasteurization, when there are heat resistant survivors was responsible for the higher final concentrations, when the storage time increased. Fruit washing, juice filtration and fruit selection, respectively, are the main responsible steps to reduce patulin levels during apple juice processing, while juice pasteurization due to high heat resistance of patulin practically does not presents effects on mycotoxin reduction in the final product. After 10000 iterations, among the 15 scenarios evaluated, only in the scenario number 1 (without mold survival to pasteurization and with low level of patulin concentration in fruits), 2 (without mold survival to pasteurization and with medium level of patulin concentration in fruits), 4 (10º/100mL of B.fulva IOC 4518 spores survival to pasteurization, with storage temperature at 21ºC and low level of patulin in fruits) and 7 (10º/100mL of B.fulva IOC 4518 spores survival to pasteurization, with storage temperature at 30ºC and low level of patulin in fruits) the limit of 50ppb of patulin would not be exceeded for any of iterations, considering medium and maximum values of the final quantity of this mycotoxin. The final distribuition of patulin concentration in apple juice best fitted to Betageneral, Lognormal and Inverse Gaussian depending on the concentration on fruit reception, with the mean and the most part of data skewed to the left / Mestrado / Mestre em Ciência de Alimentos

A risk assessment study is conducted in a denim manufacturing plant in Turkey. The study is carried out within the framework of a project on adopting the Integrated Pollution Prevention and Control (IPPC) Directive of the European Union. The scope of the assessment is fire or explosion risk with regards to hazardous chemicals present in the plant. The receptor of the study is defined as &ldquo / people&rdquo / which include the employees in the plant, employees of nearby plants and people in residential around the mill. A semi-quantitative risk assessment is carried out using checklist, a risk matrix and risk evaluation forms. The highest risks in the plant are identified as dust explosions, natural gas jet fires, natural gas explosions. Also, it is identified that due to several causes, in case of a fire or explosion the scale of an accident may enlarge instantaneously. The main warehouse is determined to carry the highest risk value in the plant. Mathematical modelling studies are conducted to calculate the hazard radius for dust explosions and natural gas fire and explosion. According to the results of mathematical modelling, the highest consequences could lead to destruction of buildings or severe injuries/fatalities of people within large hazard radius up to 700 m. The risk present at the manufacturing mill is communicated to the facility management throughout the study. Several suggestions are proposed to the facility management and some of them are already implemented.

Dans ce travail, un modèle d'appréciation quantitative des risques microbiens a été développé pour proposer des mesures de maitrise sanitaire et les relier à des métriques du risque. A partir d'observations réalisées dans un hôpital, nous avons pu caractériser le procédé de fabrication et de distribution de deux plats de boeuf en sauce, de manière à prédire l'évolution du danger Clostridium perfringens tout au long de ce procédé en utilisant les outils de la microbiologie prévisionnelle. Dans le modèle d'exposition, nous avons construit un modèle décrivant la germination puis la croissance de Clostri- dium perfringens pendant deux étapes du procédé, et un modèle décrivant la destruction thermique finale des cellules végétatives de Clostridium perfrin- gens. Ensuite, pour estimer un risque de diarrhée lié à la consommation de ces plats, nous avons construit un modèle dose réponse. De manière à pouvoir prédire la gamme des situations possibles, nous avons cherché à décrire, dans chacun des modèles, la variabilité de la réponse biologique modélisée en utilisant des données issues de méta analyse, obtenues dans des conditions variées. L'utilisation de l'inférence bayésienne sur des données issues de méta analyse nous a permis de décrire séparément variabilité et incertitude, suivant les recommandations des instances internationales. Des simulations de Monte Carlo à deux dimensions sur l'ensemble du procédé nous ont ensuite permis d'estimer le risque moyen de maladie lié à la consommation des plats et d'estimer l'incertitude sur ce risque moyen. Ces simulations nous ont aussi permis d'identifier deux étapes importantes pour la maitrise du risque et de proposer, pour chacune de ces étapes, des mesures de maitrise sanitaire, dont nous avons ensuite mesuré l'efficacité. / In this study, a quantitative risk assessment model was developed in order to propose control measures making it possible to reach some `risk based' targets. A survey, carried out in an hospital, made it possible to describe a food chain, from raw material until consumption, of two beef-in-sauce products, such as to predict the evolution of Clostridium perfringens along the whole process using predictive microbiology tools. In the exposure assessment model, we described Clostridium perfringens germination, outgrowth and growth in a first model, applied on two process steps, and final thermal inactivation of Clostridium perfringens vegetative cells in a second model. Then, in order to estimate the risk of diarrhea after the consumption of the two products, we established a dose response model. In order to predict in our model the full range of possible outcomes, we aimed at describing, in each proposed model, the variability on the modelled biological response using meta analysis data, obtained on various experimental conditions. Performing Bayesian inference on these published data, we separated variability and uncertainty, according to recommendations of international organizations. Second order Monte Carlo simulations on the whole process made it possible to estimate the mean risk of diarrhea after to the consumption of the products and to estimate the uncertainty on this median risk. Based on these simulations, two key steps for food safety were then identified in the process, leading us to propose, for each step, control measures and to measure their efficacy.