Developing and evaluating distributions for probabilistic human exposure assessments

Maddalena, Randy L. ; McKone, Thomas E.

Lawrence Berkeley National Laboratory

关键词: Us Superfund;    Us Epa;    Recommendations;    32 Energy Conservation, Consumption, And Utilization;    Water;   

DOI  :  10.2172/803868
RP-ID  :  LBNL--51492
RP-ID  :  AC03-76SF00098
RP-ID  :  803868

美国|英语

来源: UNT Digital Library

【 摘 要 】

This report describes research carried out at the Lawrence Berkeley National Laboratory (LBNL) to assist the U. S. Environmental Protection Agency (EPA) in developing a consistent yet flexible approach for evaluating the inputs to probabilistic risk assessments. The U.S. EPA Office of Emergency and Remedial Response (OERR) recently released Volume 3 Part A of Risk Assessment Guidance for Superfund (RAGS), as an update to the existing two-volume set of RAGS. The update provides policy and technical guidance on performing probabilistic risk assessment (PRA). Consequently, EPA risk managers and decision-makers need to review and evaluate the adequacy of PRAs for supporting regulatory decisions. A critical part of evaluating a PRA is the problem of evaluating or judging the adequacy of input distributions PRA. Although the overarching theme of this report is the need to improve the ease and consistency of the regulatory review process, the specific objectives are presented in two parts. The objective of Part 1 is to develop a consistent yet flexible process for evaluating distributions in a PRA by identifying the critical attributes of an exposure factor distribution and discussing how these attributes relate to the task-specific adequacy of the input. This objective is carried out with emphasis on the perspective of a risk manager or decision-maker. The proposed evaluation procedure provides consistency to the review process without a loss of flexibility. As a result, the approach described in Part 1 provides an opportunity to apply a single review framework for all EPA regions and yet provide the regional risk manager with the flexibility to deal with site- and case-specific issues in the PRA process. However, as the number of inputs to a PRA increases, so does the complexity of the process for calculating, communicating and managing risk. As a result, there is increasing effort required of both the risk professionals performing the analysis and the risk manager reviewing it. For deterministic risk assessments, the use of default inputs has improved the ease and the consistency of both performing and reviewing assessments. By analogy, it is expected that similar advantage will be seen in the field of probabilistic risk assessment through the introduction of default distributions. In Part 2 of this report, we consider when a default distribution might be appropriate for use in PRA and work towards development of recommended task-specific distributions for several frequently used exposure factors. An approach that we develop using body weight and exposure duration as case studies offers a transparent way for developing task-specific exposure factor distributions. A third case study using water intake highlights the need for further study aimed at improving the relevance of ''short-term'' data before recommendations on task-specific distributions of water intake can be made.