12.07.2015 Views

Final Program - Society for Risk Analysis

Final Program - Society for Risk Analysis

Final Program - Society for Risk Analysis

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

are being used to collect real time in<strong>for</strong>mation together with manual inspection ofbridge systems. Despite the mandate of the National Bridge Inspection Standards byFHWA, many bridges receive insufficient inspection due to limitations of funding,equipments, manpower, and the reliability of current sensing technology. A bridge is asystem with many interdependent functional elements, and the purpose of inspectionis to discover evidences of deterioration of individual elements in a timely fashionand to evaluate the consequences. Assuming that the underlying condition of eachindividual element is not directly observable, we treat inspection as a detection processwith considerable uncertainties during the assessment process. Fault tree analysisis used to provide a link between the evolving deterioration conditions of each elementof the bridge and a specific failure mode‚Äîa process that requires accuracy andreliability of the sensing and inspection system, which can be quantified by identifyingtime-critical elements in minimum cut sets. A systemic approach is developed tointegrate multiple models <strong>for</strong> the ultimate purpose of developing risk managementstrategies <strong>for</strong> improving inspection and monitoring, and <strong>for</strong> an effective resource allocationprocess. The developed methodology is expected to help bridge owners toefficiently prioritize and plan <strong>for</strong> inspection, maintenance, and remediation activitiesto reduce the risk of bridge failure.M4-I.1 Guvenc U, Small MJ, Morgan MG; umitguvenc@gmail.comCarnegie Mellon UniversityALTERNATIVE METHODS FOR AGGREGATION OF EXPERT JUDG-MENTSUncertainty is a key feature of many important technical assessments involvinghealth and safety risks. This is particularly true <strong>for</strong> models of large-scale, complex,multidisciplinary systems with many unknown elements (e.g., models <strong>for</strong> climatechange). Experts’ judgments regarding key model and related risk parameters are notalways fully aligned with each other. While it may be argued that characterizing thevariation across experts is more important than determining a single “best estimate”,such estimates do have utility when considered in the proper context. For these applicationsdifferent methods have been proposed <strong>for</strong> aggregating expert judgments,typically involving linear weights. These methods use special measures to evaluatethe “quality” of experts and the “appropriateness” of the way their opinions areexpressed, and as a result, the weighting schemes applied to expert judgments differdepending on the aggregation method used. All aggregation methods have advantagesand disadvantages under different circumstances. This paper uses simulation tocompare three popular approaches <strong>for</strong> expert judgment aggregation: likelihood method(equivalent to Bayesian weighting when experts are assumed to have equal priorweights), “classical” method, and equal weights. Both the likelihood and the “classical”method determine weights by evaluating the consistency of each expert’s judgmentswith the “observed” evidence. This is done using a set of “seed” questions to110evaluate the per<strong>for</strong>mance of the experts. The “classical” method is non-parametric,comparing the consistency of the experts’ uncertainty quantiles with the observedvalues, applied through a scoring rule developed by Cooke et al. The likelihood methodis parametric, requiring an assumed or expert-elicited probability distribution function<strong>for</strong> their estimation error. This paper evaluates these approaches under differentcircumstances and discusses the implications <strong>for</strong> characterizing uncertainty.T3-C.4 Haber LH, Kroner OL; haber@tera.orgTERAWHERE THE RUBBER MEETS THE ROAD: A PRACTICAL METH-ODS COMPENDIUM FOR RISK ASSESSORSA wide variety of dose-response approaches exist that apply increasingly datain<strong>for</strong>medmethods and can be used to address a range of problem <strong>for</strong>mulations.However, many risk assessors may not be aware of the range of tools that are availableto address specific needs and questions. To aid in communicating the methodsthat can be used to address different issues, we developed a framework, building offof the framework in the NAS (2009) report, to organize risk assessment methodsand guidances in a practical methods compendium. This internet-based frameworkaddresses qualitative and quantitative screening approaches and in-depth assessmentmethods, and includes active links to additional resources. Application of the methodsto address specific issues raised by the NAS (2009) report will be introduced, andaddressed in greater detail in the remaining talks in the symposium.P.63 Haber LT, Kaden DA, Meek ME, Schroeder J; haber@tera.orgTERA, ENVIRON, University of Ottawa, Ontario Ministry of the EnvironmentREVIEW OF ISSUES RELEVANT TO AMBIENT AIR QUALITY CRITE-RIAThe Ontario Ministry of the Environment is reviewing the science supportingair quality criteria development as part of an ef<strong>for</strong>t to document its process andenhance clarity and consistency <strong>for</strong> stakeholders. The review is also intended to identifyopportunities to refine the methods <strong>for</strong> developing ambient air quality criteria.Areas of particular interest include 1) factors affecting the choice of averaging times<strong>for</strong> sampling and setting of air standards; 2) definitions and applications of chronic,subchronic, and acute exposure and effects; 3) uncertainty factors and the interplaybetween uncertainty factors and the determination of the point of depature; 4) allocationof exposure across routes and consideration of combined exposures; and5) linear vs. non-linear/threshold vs. non-threshold dose response. In support ofthis ef<strong>for</strong>t, we reviewed the approaches used by a variety of jurisdictions in Canada,and US Federal and State agencies, as well as international and national authoritativebodies to identify approaches, issues, and best practices. Areas of controversy withinthe scientific community and areas needing additional research were also identified.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!