All three authors are AIChE Fellows and CCPS Fellows
Editorial, Letter to the Editor, and Author's Response
To the editors of Process Safety Progress
Article first published online: 16 APR 2012
Copyright © 2012 American Institute of Chemical Engineers (AIChE)
Process Safety Progress
Volume 31, Issue 2, pages 97–98, June 2012
How to Cite
- Issue published online: 17 MAY 2012
- Article first published online: 16 APR 2012
Vol. 31, Issue 1, 6–8, Article first published online: 2 AUG 2011
We are writing to express our concerns with the paper, “New Risk Acceptance Criteria for Process Safety,” that appeared in the March 2012 edition of Process Safety Progress (PSP) (Vol. 31, No. 1, pp. 6–8, DOI 10392) authored by James Moseman. In our opinion, this paper contains a number of conceptual and factual errors, of which readers should be made aware.
The paper states in Table 2 that the American Institute of Chemical Engineers (AIChE) Center for Chemical Process Safety (CCPS) has established a risk acceptance criteria of 1 × 10−5/year (presumably, the intended units are fatalities/year). No source is provided to support this value. It appears that the author referred to Ref. 5, the CCPS Guidelinesfor Hazard Evaluation Procedures, 2nd edition (HEP Guidelines). The first, second, and third editions of the HEP Guidelines provide guidance on the conduct of process hazard analyses, not risk assessments. These guidelines do not propose any values for risk acceptance criteria.
The paper also references the CCPS' Guidelines for Chemical Process Quantitative Risk Analysis – 1989 Edition (CPQRA Guidelines). Both the 1989 and 2000 editions of the CPQRA Guidelines describe methodologies for conducting risk assessments; however, neither of them proposes risk acceptance criteria. The committees that developed these guidelines recognized that there was no agreement within industry as to what might constitute appropriate risk acceptance criteria, nor was there basis for arriving at a consensus.
Risk acceptance criteria can be useful in making risk management decisions. Recognizing the usefulness of risk criteria, CCPS published in 2009 Guidelines for Developing Quantitative Safety Risk Criteria (Risk Tolerance Criteria or RTC Guidelines). This publication is not referenced in the author's paper. The RTC Guidelines were developed with the oversight of a diverse committee of industry risk and safety experts. Hundreds of data sources were reviewed and relevant risk acceptance criteria are summarized in the book. However, the RTC Guidelines, do not propose the adoption of any specific risk acceptance criteria.
One or more of us had direct involvement in the development of each of these guidelines. More broadly, based upon our collective knowledge of the CCPS publications catalog, we are confident that CCPS has never proposed any specific risk acceptance criteria. As the introduction to the RTC Guidelines notes, “The recommendation of specific risk criteria is beyond the charter of CCPS' mission.”
The risk acceptance criteria presented in Moseman's paper should be interpreted with great caution. The paper leaves considerable uncertainty as to the particular groups to which the criteria are intended to apply (e.g., workers, public, or vulnerable populations) and the level of harm addressed by the criteria (e.g., fatality or serious injury). It is also unclear as to whether the proposed criteria represent an annual frequency or a fixed probability of injury or death. The second edition of the CPQRA Guidelines presents 14 quantitatively different risk measures, all derived from the same set of incident, likelihood, and consequence data, in Table 4.11. The CPQRA Guidelines go on to state “When comparing the risk of facilities or design options, it is essential that the risks are calculated on the same basis for the comparison to be meaningful.” The paper does not clearly indicate which risk measure is to be used in conjunction with the proposed acceptance criteria. For example, are the proposed criteria to be applied to: (1) the risk to the individual most exposed to process hazards; (2) the average risk to the exposed worker population; or (3) the average risk to the total site population, which would include some workers who are not exposed to process risks?
Depending upon how the reader interprets the application of the criteria proposed in the paper, the criteria are more stringent, sometimes by several orders of magnitude, than precedents documented in the RTC Guidelines. For example, the executive summary of the paper proposes “acceptable risks (1 × 10−7), where no actions are recommended [and] unacceptable or marginally acceptable levels (1 × 10−5), where mitigation actions are recommended.” We assume that these criteria are intended to be annual risks of death for workers. For comparison, as noted in the RTC Guidelines, the UK Health and Safety Executive has established a worker risk acceptance criterion of 1 × 10−3 fatalities/year (maximum) and 1 × 10−6 fatalities/year (de minimis).
The paper asserts that a criterion of 1 × 10−7/year is not unreasonable, since this equates to the average annual risk of a person being struck by lightning. Also, the paper seeks to justify the 1 × 10−7/year proposed criterion based upon the fact that some might seek to avoid air travel “even for short trips of 100 miles when the total risk per flight is 1 × 10−7.” We are not aware of any precedent for establishing acceptance criteria for process-related risks for either workers or off-site populations based upon lightning fatality statistics or airline flight crash statistics.
For calculated frequencies below 1 × 10−4 events/year, risk may be dominated by common cause failures, human factors, and external events, which are difficult to quantitatively address in a risk assessment. The lower the frequency of hardware failures, the more important these other concerns become. Calculated frequencies below 1 × 10−6 events/year must be examined carefully due to these concerns. To ensure the credibility of the results at such low risk levels, organizations often have detailed requirements for calculating the frequency and consequences of hypothetical events before comparing the results with the organization's risk criteria.
Any reader tasked to develop risk acceptance criteria is urged to seek the guidance provided in the RTC Guidelines, which summarizes recent thinking of a large number of regulatory bodies and expert risk analysts. The RTC Guidelines also provide a more complete, and accurate, treatment of some concepts discussed in the subject paper. For example, the RTC Guidelines provide a basis for understanding that application of the “As Low As Reasonably Practicable” principle should not lead to the “arbitrary and capricious” decisions about risk acceptability asserted in the subject paper.
Papers describing RTC Guidelines and discussing the topic of risk tolerance criteria have been published recently in Process Safety Progress:
“Choosing Appropriate Quantitative Safety Risk Criteria: Applications From the New CCPS Guidelines,” (PSP, Vol. 29, Issue 4, pp. 293–298, December 2010)
“Challenges In Developing and Implementing Safety Risk Tolerance Criteria,” (PSP, Vol. 30, Issue 3, pp. 232–239, September 2011)