Supplemental Analysis to Support Postulated Events in Process Hazards Analysis for the HEAF

Lambert, H ; Johnson, G

Lawrence Livermore National Laboratory

关键词: Probability;    42 Engineering;    Detonators;    Failure Mode Analysis;    Accidents;   

DOI  :  10.2172/15002507
RP-ID  :  UCRL-ID-150819
RP-ID  :  W-7405-ENG-48
RP-ID  :  15002507

美国|英语

来源: UNT Digital Library

【 摘 要 】

The purpose of this report is to conduct a limit scope risk assessment by generating event trees for the accident scenarios described in table 4-2 of the HEAF SAR, ref 1. Table 4-2 lists the postulated event/scenario descriptions for non-industrial hazards for HEAF. The event tree analysis decomposes accident scenarios into basic causes that appear as branches on the event tree. Bold downward branches indicate paths leading to the accident. The basic causes include conditions, failure of administrative controls (procedural or human error events) or failure of engineered controls (hardware, software or equipment failure) that singly or in combination can cause an accident to occur. Event tree analysis is useful since it can display the minimum number of events to cause an accident. Event trees can address statistical dependency of events such as a sequence of human error events conducted by the same operator. In this case, dependent probabilities are used. Probabilities/frequencies are assigned to each branch. Another example of dependency would be when the same software is used to conduct separate actions such as activating a hard and soft crow bar for grounding detonator circuits. Generally, the first event considered in the event tree describes the annual frequency at which a specific operation is conducted and probabilities are assigned to the remaining branches. An exception may be when the first event represents a condition, then a probability is used to indicate the percentage of time the condition exists. The annual probability (frequency) of the end state leading to the accident scenario in the event tree is obtained by multiplying the branch probabilities together.