Final Program - Society for Risk Analysis

T3-D.4 LaRocca S, Guikema SD, Cole J, Sanderson E; larocca@jhu.eduJohns Hopkins UniversityBROADENING THE DISCOURSE ON INFRASTRUCTURE INTER-DEPENDENCE BY MODELING THE ‘ECOLOGY’ OF INFRASTRUC-TURE SYSTEMSInterdependencies among infrastructure systems arise for many reasons, includinggeographic proximity inducing common cause failures, direct dependence forphysical flows of information, and common maintenance and repair actions. However,existing modeling of the risk and reliability of interdependent infrastructuregenerally deals with a limited subset of these sources of dependence, often focusingonly on physical flows and geographic proximity. In this paper we show how a modelingconstruct recently proposed for ecological modeling can be used to give a broaderpicture of dependencies between infrastructure systems and system elements. Thisapproach is based on Muir webs, a modeling approach first proposed for modelingcomplex interdependencies in a pre-European ecological-human community in theU.S. by Sanderson 2009. Muir webs generalize from traditional predator-prey relationshipsto consider broader interactions such as dependencies on biotic but notdirectly consumed factors (e.g., shade for certain species of trees). They also includeinterdependencies due to abiotic factors (e.g., soil types and climate) and human ‘management’of the environment. These factors are considered through a dependencynetwork describing (1) what factors a given organism depends on and (2) what otherorganisms and factors depend on a given organism performing its role in the environment.In this paper we show how a Muir web can be used to model interdependentinfrastructure system reliability. Here, each ‘organism’ in the infrastructure Muir webis either a component of the system (e.g., a pump or valve in a water distributionsystem) or is a factor needed by some element(s) of the system for it to perform itsintended role (e.g., for a water pump: stable soil, a water supply, and proper maintenance).We use this expanded representation of the dependencies and interdependenciesand demonstrate how to estimate system reliability through a simulation-basedapproach.M3-H.3 Lathrop JF, Post JM; jlathrop@innovativedecisions.comInnovative Decisions, Inc. and Political Psychology Program, Elliot School of International Affairs,George Washington UniversityTHE MODELER MEETS THE EXPERT ON TERRORIST DECISIONMAKING: RISK MANAGEMENT BASED ON TWO CULTURESThis paper will be delivered as a dialog between a “Modeler” and an expert,an “SME” on terrorist decision making. Terrorism risk management offers uniquechallenges to risk management Modelers and terrorism SMEs. Those challenges callfor a new paradigm in analysis. You can’t address the problem by the modeler buildingthe model then going to the SME to populate it. The SMEs know more aboutthe necessary structures of the models than the Modelers do. Before the Modelerstarts building his model, he should elicit the architecture from the SME, starting ata narrative level, build his model upon that architecture, then populate it with SMEjudgments. This risk management problem needs the information collection and aggregationstructure of PRA, but there needs to be an analytic strategy surroundingPRA to make full use of the information from the SMEs to advise Blue risk management.The Modeler and the SME come from two different cultures, but they have towork together: The SME needs the Modeler to transform his knowledge into formsuseful for risk management advice; The Modeler needs the SME first to provide himhis model architecture, then to provide the data with which to populate the model.Four fundamental considerations: 1.) Don’t Fight the Last War: as if Red behavioris predictable by a simple equation, as in the Cold War, and don’t key on currentlyknown Reds (e.g. al Qaeda, McVeigh) when the bulk of the threat is future Reds. 2.)Don’t Play the Wrong Game, examples: don’t play the game at the defend-each-targetlevel if Red is playing at a higher level, and don’t play as if Red maximizes fatalities ifRed is maximizing terror. 3.) Avoid Mirror Imaging: assuming Reds behave/decide ina linear, rational, “Modeler’s” manner. 4.) Maintain Epistemological Modesty: Avoidany false precision of a model result, which could e.g. over-focus on defending highscoringcities. This presentation will end with suggested strategies for addressing thechallenges presented.M4-E.3 Lawrence R, Brown SM; Susan_Brown@McCormick.comMcCormick & Company IncIDENTIFICATION, MONITORING AND MANAGEMENT OF RISKSIN THE SPICE INDUSTRYThe American Spice Trade Association published in the spring of 2011 a guidancedocument aimed at reducing the risk for contamination of spices with Salmonellaand other pathogens. The guidance contains five key recommendations. 1)Minimize risk for introduction of filth throughout the supply chain 2) Prevent environmentalcontamination, cross-contamination, and post-processing contaminationduring processing and storage 3) Use validated microbial reduction techniques 4)Perform post-treatment testing to verify a safe product 5) Test to verify a clean andwholesome manufacturing environment The recommendations where based on thebest practices from the industry. Hear how one firm, McCormick & Company, hasbeen applying these as part of their on-going commitment to food safety. McCormickhas been in the business of sourcing ingredients globally since 1889. Few companieshave their know-how and experience in sourcing pure, wholesome and safefood products from around the world. Join McCormick as they share a case study ofthe company’s experiences and strategies for developing programs that meet the importchallenges of today. Learn how it successfully manages the sourcing of materialfrom China, India, Indonesia and other developing nations that have historically beenchallenged to meet the standards of good agricultural and manufacturing practices.133