(55m) Human Reliability Analysis of Pre-Initiator and Initiator Events - What Methodological Adjustments Must be Adopted in Petro-Hra to Analyze Pre-Accident Operator Actions?

The adverse impacts of human error have been demonstrated by major accidents in different industrial segments, which in turn reveals the need for a means of properly assessing and mitigating the risks associated with human failure events in complex socio-technical systems [1]. These are the primary goals of a Human Reliability Analysis (HRA), achieved by its main functions, which can be summarized as:

- identifying what human errors can occur while individuals are carrying out a task;

- identifying the contextual factors that influence human performance and may represent error traps, the so-called Performance Shaping Factors (PSFs);

- deciding how likely these errors can be, by systematically estimating their numerical human error probabilities;

- recommending measures capable of acting on the identified error producing conditions to reduce the likelihood of human errors.

In HRA methods, the likelihood of a human error is estimated by using a nominal human error probability (NHEP) value that is then adjusted to reflect the influence of contextual factors that specifically decrements or improves human performance, thus respectively increasing or decreasing the human error probability. Following this rationale, the main elements for quantifying the probability of a human failure event are the NHEP and the adjustment factors (or multipliers) that represent the level of influence that each of the human performance shaping factors considered in a human reliability analysis has.

The NHEP is supposed to contain all small influences that can contribute to errors that are not covered by the performance shaping factors. The PSFs include individual characteristics of those involved in the task, besides aspects related to the work environment, the organization, or the task itself, encompassing items such as experience, level of training, task complexity, quality and availability of procedures, time available to perform the task etc.

Among the existent methods of human reliability analysis, the Petro-HRA was specifically developed for the petroleum industry [2, 3]. The Petro-HRA was adapted from a method which originated in the nuclear industry, the so-called SPAR-H (Standardized Plant Analysis Risk – Human Reliability Analysis) [4]. The main adaptations performed to make the method suitable for application in the petroleum industry included:

- changes in the definitions of the Performance Shaping Factors, their levels of influence and correspondent multipliers [3]. As a result, nine PSFs were adopted in the Petro-HRA, which are: time, threat stress, task complexity, experience and training, procedures, human-machine interface (HMI), attitudes to safety, work and management support, teamwork, and physical working environment.

- the adoption of the diagnosis NHEP of 0.01 from SPAR-H for all tasks in Petro-HRA. The 0.01 value means that a task step fails 1 out of 100 times (or opportunities for failure). Although the SPAR-H method distinguishes between two task types (diagnosis and action) and provides a separate NHEP for each (0.01 and 0.001, respectively), it was considered that all task steps include a combination of diagnosis and action, and that most of the tasks performed in an accident sequence involve a large cognitive component [2]. As a result, only the diagnosis NHEP was adopted in the Petro-HRA method.

At this point, it is worth explaining that the Petro-HRA was primarily developed to analyze post-initiator events, meaning the human failures that occur as part of human responses to abnormal situations and that are required to prevent or reduce losses in an accident sequence. These events have indeed a large cognitive component and were prioritized because they represent human failure events frequently modelled in Quantitative Risk Analyses (QRAs). Examples include human actions performed in a control room in response to leakage scenarios or other loss incidents. Although the Petro-HRA is a method for both qualitative and quantitative assessment of human reliability, the method was mainly developed to provide input for QRAs, which explains the focus on post-initiator events.

More recently, a separate guideline was issued for Analysis of Pre-Accident Operator Actions (APOA) [5]. This new method was developed to complement the Petro-HRA, providing a means for assessing the human reliability of pre-initiator and initiator events. While pre-initiator events are those human errors that introduce (or fail to reveal) latent failures during inspection, maintenance and test activities in safety critical equipment, initiator events comprise human errors that start an accident sequence, solely or in combination with technical failures. Nonetheless, the APOA method is purely qualitative and, as a result, let the HRA analysts with no means for estimating a numerical human error probability for these pre-accident events. These HEP values may find application in LOPA (Layer of Protection Analysis) so that more realistic frequencies of initiating events can be adopted, rather than using database values which rely on assumptions that not always represent the contextual factors and potential error producing conditions in place. Additionally, quantifying the likelihood of pre-accident operator actions may be a regulatory requirement, with the Brazilian offshore oil and gas regulation [6] being an example.

The fact that the Petro-HRA method prioritized post-initiator events does not mean that it cannot be applied for human reliability analysis of pre-accident operator actions, provided that some methodological adjustments are taken into consideration. It should be mentioned that the SPAR-H, the nuclear HRA method from which Petro-HRA has its origins, does not present this sort of limitation, and can be applied to pre-initiator, initiator, and post-initiator actions. Hence, this paper aims to explore the methodological adjustments required to perform a proper human reliability analysis of pre-accident operator actions with Petro-HRA.

A clear distinction between diagnosis and action tasks will be established to identify the cases where the less conservative action NHEP of 0.001 should be applied. As a result, examples of pre-initiator and initiator human failure action events that are dominated by rules and skills and hence, do not involve a high cognitive demand will be discussed. Additionally, an assessment will be performed to verify the need to consider additional PSFs, other than the nine factors established in the Petro-HRA. Moreover, a discussion on the most proper method for modelling pre-accident human errors will be performed. The Operator Action Events Trees (OAETs) used in Petro-HRA may not be fit for purpose to model pre-initiator and initiator events, so that other methods, such as Fault Tree Analysis (FTA), may represent a better approach to model how a human error (or combination of failures) can produce an undesired event. With these discussions, the authors intend to fill a gap in HRA for the petroleum industry, expanding the application of the quantitative part of the Petro-HRA method to pre-initiator and initiator events.

References:

[1] Kirwan, B. (1994). A Guide to Practical Reliability Assessment. Taylor and Francis Group.

[2] Blackett, C., Farbrot, J. E., Øie, S., & Fernander, M. (2022). The Petro-HRA Guideline. Rev. 1. Volume 1. Institute for Energy Technology (IFE). Available at <https://ife.brage.unit.no/ife-xmlui/bitstream/handle/11250/2989743/The%2bPetro-HRA%2bGuideline%252C%2bRev.1%252C%2bVol.1.pdf?sequence=1&isAllowed=y> (accessed on the 7^th of November 2023).

[3] Blackett, C., Farbrot, J. E., Øie, S., & Fernander, M. (2022). The Petro-HRA Guideline. Rev. 1. Volume 2. Institute for Energy Technology (IFE). Available at <https://ife.brage.unit.no/ife-xmlui/bitstream/handle/11250/2989744/The%2bPetro-HRA%2bGuideline%252C%2bRev.1%252C%2bVol.2.pdf?sequence=1&isAllowed=y> (accessed on the 7^th of November 2023).

[4] Gertman, D., Blackman, H., Marble, J., Byers, J., & Smith, C. (2005). The SPAR-H Human Reliability Analysis Method. NUREG/CR-6883. Washington, DC, U.S. Nuclear Regulatory Commission.

[5] Øie, S., & Fernander, M. (2023). Analysis of Pre-Accident Operator Actions (APOA). DNV. Available at < https://www.dnv.com/Publications/analysis-of-pre-accident-operator-actions-247857> (accessed on the 7^th of November 2023).

[6] ANP (The Brazilian National Agency for Petroleum, Natural Gas and Biofuels). 2007. The Technical Regulation of the Operational Safety Management System of Maritime Oil and Natural Gas Drilling and Production Facilities (SGSO). ANP Resolution N°43/2007.