(147a) Proven Approaches to Ensuring Operators Can Respond to Critical Process Deviations in Time (Human Response IPL)

Humans can be the cause on an accident scenario (the Initiating Event [IE]) or humans can serve or participate as an independent protection layer (IPL). Validating Human IPLs has been a show stopper for many companies considering the use of human response as an IPL. Human IPLs include preventative steps that may stop a scenario from progressing once it is initiated, but more typically the human IPLs are responses to alerts or alarms or troubling readings and sample results.

 This paper first describes the fundamentals of clear alarms, practical actions, and having enough time to perform the action, all without being in harm’s way at the end of the action. This paper builds upon earlier studies (based on similar papers from 2010and 2011)^{13, 14} of how to collect the data needed for directly measuring the probability of failure on demand (PFD) of the human response. The preferred method for data collection covers the training requirements that should be met, proof drills for response to alarms, simulations and tests, and frequency of proofs, and of course the effect of human factors on human error rates. An example is provided of how a simple data collection and validation method can be set up within a company. This paper also provides an overview of alternative methods for estimating the PFD of a Human IPL, based on plant and scenario specific factors (such as stress factors, complexity, and communication factors); and the paper evaluates and compares alternative approaches to validating human IPLs, including expert judgment based on frequent practice of trouble-shooting by operators. All of these methods were available in Appendix B and C of the initial full draft of the CCPS book, Guidelines for Initiating Events and Independent Protection Layers, April 2012 (unpublished), but unfortunately only a small fraction of one of the Appendices was retain in the version that was published in January 2015. This paper provides that missing information.