《AICHE-G-21-1995.pdf》由会员分享，可在线阅读，更多相关《AICHE-G-21-1995.pdf（48页珍藏版）》请在三一文库上搜索。
1、TOOLSFORMAKING ACUTE RISK DECISIONS with Chemical Process Safety Applications CENTER FOR CHEMICAL PROCESS SAFETY of the AMERICAN INSTITUTE OF CHEMICAL ENGINEERS 345 East 47th Street, New York, New York 10017 Copyright 1995 American Institute of Chemical Engineers 345 East 47th Street New York, New Y
2、ork 10017 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise without the prior permission of the copyright owner. Library of Congress Cataloging-in
3、 Publication Data Tools for making acute risk decisions with chemical process safety applications. p. cm. Includes bibliographical references and index. ISBN 0-8169-0557-6 1. Chemical plantsRisk assessment. I. American Institute of Chemical Engineers. Center for Chemical Process Safety. TP155.5.T65
4、1994 660 .2804dc20 94-2462 CIP PRINTED IN THE UNITED STATES OF AMERICA 6 5 4 3 2 1 99989796 9 5 This book is available at a special discount when ordered in bulk quantities. For information, contact the Center for Chemical Process Safety of the American Institute of Chemical Engineers at the address
5、 shown above. It is sincerely hoped that the information presented in this document will lead to an even more impressive safety record for the entire industry; however, the American Institute of Chemical Engineers/ its consultants/ CCPS subcommittee members/ their employers/ their employers7 officer
6、s and directors/ and Applied Decision Analysis/ Inc. disclaim making or giving any warranties or representations/ express or implied/ including with respect to fitness/ intended purpose/ use or merchantability and/or correctness or accuracy of the content of the information presented in this documen
7、t. As between (1) the American Institute of Chemical Engineers/ its consultants/ CCPS subcommittee members/ their employ- ers/ their employers officers and directors/ and Applied Decision Analysis/ Inc. and (2) the user of this document/ the user accepts any legal liability or responsibility whatsoe
8、ver for the consequence of its use or misuse. PREFACE The Center for Chemical Process Safety, first organized in 1985 as a Directorate of the American Institute of Chemical Engineers, has been responsible for the development of a significant number of Guidelines books in process safety. The first bo
9、ok in the Guidelines series, Guideline for Hazard Evaluation Proce- dures, was published in 1985. This book focuses on identifying hazards using tools that are familiar to the process industries. After achieving enormous success an updated second edition was published in 1992. In 1989, another Guide
10、lines book was published entitled Guidelines for Chemical Process Quantitative Risk Analysis (CPQRA). The CPQRA book pre- sents tools for the quantitative assessment of risk. At the time of publication, the use of quantitative tools in the process industries was much more limited than the use of sub
11、jective hazard evaluation procedures. Our book serves to extend these two previous books. It addresses tools used for decision making where risks have been assessed. This book represents a departure from the guidelines series in that some of the tools and method- ologies addressed have been applied
12、to other industry problems, but only infrequently at best to process risk decisions. Decisions in all facets of business, especially process risk decisions, have become more complex and more critical to the long term success of the industry. Better tools and methods are needed to help decision maker
13、s reach the best decisions. Risk decision making has evolved in sophistication but is still constrained by the methods and tools currently in use. Historically, the process industries have based most of their risk decisions on standards, experience and good engineering practices without the need for
14、 formal decision processes. Process risk decisions were relatively straight forward. In many cases, process risk standards were easily met. Some companies in the process industries have implemented quantitative criteria for assessment of risk tolerability, particularly with regard to the safety of t
15、heir employees. In these instances a risk is considered either tolerable or not tolerable. The criteria can often be met simply and inexpensively. How- -,-,- ever, where risk to the public is concerned, establishing a single numerical criterion is difficult if not impossible. One could specify a thr
16、eshold above which the risk is not tolerable under any circumstances. However, below this threshold it is extremely difficult to determine how safe is safe enough. Some governments are specifying thresholds for both individual and societal risk above which the risk of acute process incidents is cons
17、idered not tolerable. Comparison of one process risk with another process risk is an approach that has been tried in industry to a limited extent. There is a certain simplicity and philosophical appeal in attempting to keep all risks within a similar range. However, some risks will fall significantl
18、y out of the desired range even after available risk reduction measures are evaluated. For these risks a more thor- ough decision-making strategy or approach may be required. In addition to chemical process risk, other factors may affect the decision. Possible factors include financial cost, corpora
19、te image, employment of work- ers, and many others. Addressing each of these factors and their associated uncertainties, however, adds complexity. In many instances the alternatives are costly and represent a broad spectrum of possible options. In addition, recent changes in the regulatory environme
20、nt in which process industries must operate have made decision making even more complex. Public con- cerns, pressure from environmentally focused groups, and regulatory agen- cies may all have a bearing on the decision. Often, these concerns can conflict with one another. A consistent and logically
21、sound approach can help ensure that appropri- ate resources are made available and allocated effectively to risk reduction activities. Decision aids are tools to assist in these decisions. The process industry has limited experience in applying formal decision aids to the com- plex risk decisions it
22、 faces. This book provides a collection of decision aids that have been successfully applied to other problems such as strategic planning, R nearby points represent alternatives that the decision maker perceives as similar. Multiattribute utility function: A function that assigns a numerical measure
23、 of the utility to a value measure for each of a set of attributes of concern. Net present value (NPV): A measure of the value today of a stream of future costs and benefits, used to compare alternatives with differing streams of costs and benefits. If Rt are the benefits in year i, Ct are the costs
24、 in year t, n is the number of years, and r is the discount rate per year, the net pre- sent value is: NPV = (R0 - C0) + (Ri - Ci)/(I + r) + (R2 - C2)X(I + r)2 + (R3 - C3)X(I + r)3 + + (Rn - Cn)/(1 + r)n Nominal group technique: A process designed to help generate ideas, set priorities, and reach de
25、cisions within a group context that strives to en- sure equal contribution from all participants. Nonlinear programming: A branch of mathematical programming con- cerned with solving planning problems in which either the objective function or some of the constraints are nonlinear functions of the de
26、ci- sion variables. Objective function: In mathematical programming, the mathematical ex- pression that describes the goal (or objective) of the decision maker. Objective: An expression of a decision makers goal in terms of an object and a direction of preference (e.g., “minimize fatalities/ where f
27、atalities is the object and minimize is the direction of preference). Objectives hierarchy: A graphical representation of the relationships among the objectives of a decision maker; each level of the hierarchy shows the objectives that contribute to the broader objectives shown at higher lev- els. O
28、pportunity cost: The implied cost of an investment due to the inability to use the capital invested in a particular, different way (e.g., the opportu- nity cost of a plant expansion might include the income from another, more productive use of the required land). Outcome: The specific result of a de
29、cision and the resolution of an uncer- tainty or series of uncertainties. Parameter: A numerical quantity that is assumed to be known. Payoff matrix analysis: A decision aid that determines the expected value of distinct alternatives allowing for future uncertainty, using a simple tabu- lar form. Th
30、e problem must involve only one decision and one uncer- tainty, and the uncertainty must not depend upon the decision. Point estimate: A single number used to summarize an uncertain quantity. Portfolio analysis: A decision aid that helps a decision maker determine the combination of risky alternativ
31、es that offers the best combination of high return and low risk. Preferential dependence: Preferential dependence exists between two attrib- utes if the preference for one attribute depends upon the quantity of the second attribute. Preferential independence: Two attributes have preferential indepen
32、dence if the preference for one attribute does not depends upon the quantity of the second attribute. Probabilistic: Pertaining to the use of probability to represent uncertainty. Probabilistic dependence: Probabilistic dependence exists between two un- certainties if knowing the resolution of one u
33、ncertainty gives informa- tion about the possible resolution of the other. For example, if the probability of rain is 20% when there is no information on cloud cover, and the probability of rain when the day is cloudy is 50%, rain is depend- ent on cloudy conditions and dependence exists between rai
34、n and cloudi- ness. Probabilistic dependence may also exist between a decision and an uncertainty. If a decision changes the probabilities of the states of an un- certainty, the uncertainty is dependent on the decision. For example, de- cisions to implement safety measures reduce the probability of
35、accidents; therefore, accident uncertainty is dependent on decisions to implement safety measures. Probabilistic independence: Probabilistic independence exists between two uncertainties if the resolution of one uncertainty gives no information about the possible resolution of the other. For example
36、, if a pair of dice are rolled and then a coin is flipped, the probability of heads is assumed to be 50% regardless of the result of rolling the dice. The coin flip is prob- abilistically independent of the roll of the dice. An uncertainty can also be described as independent of a decision. For exam
37、ple, the chance of rain is independent of the decision to carry or not carry an umbrella. Probabilistic Risk Assessment (PRA): A commonly used term in the nu- clear industry to describe the quantitative evaluation of risk using prob- ability theory. Also known as Probabilistic Safety Assessment (PSA
38、). Probability: A number that expresses the likelihood of occurrence of a possi- ble state of an uncertainty. By definition, a probability must be a number between O and 1, and the sum of probabilities for all possible states of an uncertainty must be 1. Probability density function: A mathematical
39、description of the relative like- lihoods of occurrence of the possible states of an uncertainty. For a con- tinuous function, the integral of this function between two values is the probability that the true value lies in the interval between the values. For discrete variables, it is often called a
40、 probability mass function. Probability distribution: A mathematical function that associates prob- abilities (e.g., weather is rainy, cloudy, or sunny) with uncertain events (e.g., amount of rainfall). Rainbow diagram: A graphical output from a probabilistic sensitivity analy- sis used to examine t
41、he sensitivity of a decision to an input. Rate of return: The discount rate at which the net present value of a stream of costs and benefits over time is zero. Risk: A measure of economic loss, human injury, or environmental damage, in terms of both the incident likelihood and the magnitude of the l
42、oss, in- jury, or damage. Risk analysis: The development of a quantitative estimate of risk based on engineering evaluation and mathematical techniques for combining esti- mates of incident consequences and frequency. Risk assessment: A process by which the results of a risk analysis (i.e., risk est
43、imates) are prepared for use in decisions, either through relative rank- ing of risk reduction strategies or through comparison with risk criteria. Risk attitude: A decision makers preferences towards facing variation in possible losses and gains. In decision analysis, risk attitude is expressed mat
44、hematically through a utility function. Risk-averse: A description of a decision makers risk attitude in which the value for a risky alternative is lower than the expected value of the alter- native. Risk-neutral: A description of a decision makers risk attitude in which the value for a risky altern
45、ative is equal to the expected value of the risky al- ternative. Risk-preferring: A description of a decision makers risk attitude in which the value for a risky alternative is higher than the expected value of the risky alternative. Risk profile: A cumulative probability distribution over the range
46、 of values associated with all possible outcomes resulting from selecting a given al- ternative. Satisficing: A decision criterion that selects the alternative that satisfies as many of the decision makers goals as possible. Scenario: A set of states across several uncertainties. For example, two si
47、m- ple uncertainties are “how long will I wait for the bus/ and “will it rain today.“ The “how long will I wait for the bus“ states might be “10 min- utes“ and “20 minutes.“ The “will it rain today“ states might be “rain“ and “no rain.“ “Wait for the bus 20 minutes“ and “rain“ comprise one possible
48、scenario. As a second example, consider a fire hazard. Two un- certainties associated with the hazard might be the chance of a flamma- ble liquid leak and the presence of an igniting flame or spark. One scenario is that the flammable liquid leaks and there is no igniting flame or spark. An alternati
49、ve scenario is that the flammable liquid leaks and is ignited by a flame or spark. Sensitivity: The sensitivity of a measure to a parameter is defined as the change in the measure per unit change in the parameter. How much an output of a model changes with change in one or more inputs. Sensitivity analysis: A technique in which one or more parameters are var- ied to examine their impact on a measure. Shadow price: The decision makers value for a unit of scarce resource. In economics, refers to the price at which something would