An integrated method of human error likelihood assessment for shale-gas fracturing operations based on SPA and UAHP

https://doi.org/10.1016/j.psep.2019.01.003Get rights and content

Highlights

  • Shale gas fracturing is a highly industrialized process of hazardous operations.

  • An integrated method using UAHP-SPA is proposed to assess human error likelihood.

  • SPA is applied for addressing epistemic uncertainties in the latent human errors.

  • The SPA-based UAHP is used to reduce aleatory uncertainties for index weights.

  • The proposed method can effectively provide human error likelihood state and trend.

Abstract

In drilling industries, hydraulic fracturing of unconventional shale-gas wells is a highly industrialized process involving multiple equipment, substances, and operational stages. However, fracturing operations also involve extremely hazardous work relating to high-pressure chemicals, heavy equipment, and flammable gases. During the fracturing process, human error frequently causes sand blockages, equipment damage, and fracturing fluid leakage. An assessment of human error likelihood is essential for the prevention of resulting incidents, and thereby improving the safety and stability of the entire shale-gas fracturing operations. However, due to the latency, unpredictability, and variety of human errors in fracturing operations, previous studies have not sufficiently considered uncertainties. Hence, an integrated method that applies the UAHP-SPA model is proposed to assess human error likelihood in the fracturing process. The indexes of human error likelihood assessment in fracturing operations are first established based on 4M theory. Then, the Set Pair Analysis (SPA) technique is used to consider epistemic uncertainties from assessed objects, assessment procedures, assessor subjectivities, and incomplete data. Furthermore, the SPA-based Uncertainty Analytic Hierarchy Process (UAHP) is applied for reducing aleatory uncertainties in the index weight optimization. The proposed method is able to effectively provide static likelihood states and dynamic likelihood trends of various human errors. To illustrate its validity, an on-site fracturing operator is selected as a test case. Results show that this UAHP-SPA model is more accurate and practical compared to conventional approaches.

Introduction

Hydraulic fracturing is a significant technique used to improve the production of shale-gas resources, and has become increasingly popular in drilling industries (Rahm, 2011; Wang et al., 2018). It is a highly industrialized process involving multiple equipment, substances, and operational stages. As shown in Fig. 1, there are different kinds of fracturing equipment including various types of trucks: fracturing, blender, coiler, manifold, monitor, sand storage, water storage, and chemical storage. Typical fracturing operations also require fracturing fluid (i.e., water and chemicals) and fracturing proppant (i.e., sand). In addition, a life-cycle fracturing process comprises three major stages: pre-fracturing, fracturing, and post-fracturing. The first stage mainly includes equipment placement and connection, start-up checks, pumping and evacuation, pressure testing, and perforation. The second stage also consists of many sub-stages, such as fluid displacement, packer sealing, fluid extrusion, sand addition, displacement and extrusion, and pump and well shutdown. Lastly, the post-fracturing stage encompasses fluid purging, equipment disassembly, truck evacuation, and fluid flow-back.

Fracturing operations entail extremely hazardous work involving high-pressure chemicals, heavy equipment, and flammable gases (Engelder and Zevenbergen, 2018). Human errors frequently result in sand blockages, equipment damage, and fracturing fluid leakage, which can cause serious water pollution, major casualties, and heavy equipment accidents. Consequently, high-risk events can trigger a domino effect and cause the abandonment of entire shale-gas wells.

The US Environmental Protection Agency (U. S. Environmental Protection Agency, 2010) suggested that an assessment of the relative influences of human errors would contribute to improving the overall safety of shale-gas fracturing. The U. S. Environmental Protection Agency, (2015) and Clancy et al. (2018) further investigated the potential causes of produced water spills and found that most of them were caused by human errors. In addition, Time magazine (US) reported that a blowout of the Chesapeake Energy gas well in northeastern Pennsylvania that resulted in the spillage of gallons of fracturing fluid on the surrounding ground; a subsequent accident survey indicated one of the main reasons for the blowout to be human error (Walsh et al., 2011). Finally, Heinberg et al., (2014) concluded that European fracturing companies seeking to improve operational safety should gain experience in the elimination of human errors from North American industry.

The shale-gas fracturing industry in China is facing huge challenges as it is still in the start-up and growth stages. Firstly, most of the operators are chosen from conventional oil and gas well fracturing teams but only receive insufficiently basic, short-term training prior to commencing in their roles. This results in a severe shortage of operational experience, professional knowledge, and safety skills, with human errors frequently occurring over a whole fracturing period. Secondly, shale-gas wells are mostly located in the mountainous areas of southeast China and are characterized by having small areas, requiring the use of hazardous materials, and harsh environments (Zhai et al., 2012; Dong et al., 2012; Li et al., 2013). Additionally, fracturing operations are monotonous, of long duration, and require high physical strength. Due to the challenging operating conditions, fracturing operators often suffer psychological and physical fatigue, which significantly increase the likelihood of human errors. Finally, different types of human errors, such as partial observations, incorrect judgment, and nonscientific decisions, frequently occur during the fracturing process; the latent, unpredictable, and various natures of human errors make them difficult to study. An assessment of human error likelihood is required to prevent accidents and ensure the overall safety and stability of fracturing operations.

There is a rich breadth of published studies associated with human reliability analysis or human error likelihood assessment in oil and gas industry. One research group led by Dr. Faisal Khan proposed some methods for offshore / marine operations, facilities, systems, emergencies, or evacuations, such as Success Likelihood Index Method (SLIM) (DiMattia et al., 2005; Khan et al., 2006; Abbassi et al., 2015; Islam et al., 2017a), Human Error Assessment and Reduction Technique (HEART) (Noroozi et al., 2014a, 2014b; Islam et al., 2017b), Technique of Human Error Rate Prediction (THERP) (Abbassi et al., 2015), Bayesian Network (BN) (Musharraf et al., 2013; Islam et al., 2018a, 2018b), Risk Matrix (RM) (Deacon et al., 2010, 2013), and so on. Another research group led by Dr. Emre Akyuz also developed new techniques for offshore units, marine accidents, as well as maritime ships and tankers, including Human Factor Analysis and Classification System (HFACS) (Akyuz and Celik, 2014), ANP-based HFACS (Akyuz, 2017), Cognitive Reliability and Error Analysis Method (CREAM) (Akyuz and Celik, 2015a; Akyuz, 2015), HEART (Akyuz and Celik, 2015b), AHP-based HEART (Akyuz and Celik, 2015c, 2016a), fuzzy HEART (Akyuz and Celik, 2016b), and fuzzy SLIM (Akyuz, 2016). Similarly, other studies used these methodologies. Ung (2015, 2018) firstly put forward a new fuzzy CREAM model for maritime human reliability analysis, and then integrated three conventional methods – fuzzy CREAM, BN, and fault tree analysis – for human error assessment in oil tanker grounding. Based on BN, Strand and Lundteigen (2016, 2017) incorporated human factors modelling and human-machine interface in human reliability analysis of offshore drilling operations. Cai et al. (2013) applied dynamic BNs to assess human factor barrier failure on offshore blowouts, where human factors were classified into individual, organization, and group. Using fuzzy HEART, Kumar et al. (2017) developed an approach to quantify human error probabilities and selected an LPG refueling station to illustrate its validity. Zhou et al. (2017) studied human factor risks in oil and gas drilling industries wherein human errors were identified in detail and their likelihoods and severity levels were calculated to establish their RM. Theophilus et al. (2017) presented a novel HFACS that was suitable for catastrophic oil and gas accidents. In sum, only fuzzy HEART, fuzzy SLIM, fuzzy CREAM, and BN considered the uncertainties.

However, due to the latency, unpredictability, and variety of human errors in fracturing operations, these existing studies are not adequate as they do not overcome epistemic uncertainties (i.e., fuzziness, grayness, and incompleteness). In comparison, Set Pair Analysis (SPA) has two main advantages. It can comprehensively address epistemic uncertainties from assessed objects, assessment procedures, assessor subjectivities, and incomplete data (Zhao, 1992; Zhou and Zhang, 2013); it is considered as the combination of certainties and uncertainties because it uses the identity-discrepancy-contrary model, which is more applicable to decision-making and practical conditions (Li et al., 2016; Yan and Xu, 2018). Furthermore, SPA can provide the static states and dynamic trends of the assessed objects to decision-makers (Yue et al., 2014; Chong et al., 2017). Such results contribute to the reduction of high-risk factors and can be incorporated in operational guidelines and safety checklists. At present, SPA has been widely applied in many fields, such as environmental pollutions (Yue et al., 2014; Yang et al., 2014; Li et al., 2016), occupational hazards (Chong et al., 2017), and facility failures (Zhang and Zhang, 2010; Wang et al., 2015; Yan and Xu, 2018). Particularly in oil and gas industries, some studies introduced the model of SPA to safety management of oil depot (Zheng and Chen, 2006), risk evaluation of urban gas pipeline system (Zhou and Zhang, 2013), failure possibility assessment of LNG terminal (Qi et al., 2014), and failure probability analysis of submarine pipeline (Shu et al., 2017). Hence SPA is selected as the basic technique of human error likelihood assessment in this paper.

The top priority in SPA is how to establish assessment indexes. Current studies mostly apply performance shaping factors (PSFs) and performance influencing factors (PIFs) for human error analysis (Kim and Jung, 2003; Dalijono et al., 2006; Landucci and Paltrinieri, 2016; Rasmussen and Laumann, 2018). However, these two methods can be assessed by relying on expert opinions. As there is currently little available information on human errors in shale-gas fracturing operations, it is challenging to use PSFs and PIFs in this paper. 4 M theory is one of the typical accident causation theories in systems engineering (Li, 2009; Yuan, 2015). From a system-wide aspect, it can reasonably, comprehensively, and effectively analyze human errors in the complex fracturing process. It is also clear and concise, and accessible to professionals or laymen. Therefore, the indexes of human error likelihood assessment are established based on the 4 M theory for SPA.

Traditionally, the SPA weight matrix is established based on Analytic Hierarchy Process (AHP) (Zheng and Chen, 2006; Zhou and Zhang, 2013; Chong et al., 2017). AHP only quantifies the uncertainties using a certain number from one to nine. However, assessors cannot accurately make pairwise comparisons for assessment indexes due to the uncertainties of human errors in fracturing operations. Therefore, the Uncertain Analytic Hierarchy Process (UAHP) is applied for weights optimization using interval numbers to reduce aleatory uncertainties and improve the accuracy of index weights (Wei et al., 1994; Wu and Zhu, 2002). In addition, due to the inconvenience of interval weights, SPA is specifically used to turn them into crisp values.

Therefore, this paper proposes an integrated method of human error likelihood assessment for shale-gas fracturing operations based on a UAHP-SPA model. The rest of the paper is organized as follows. Section 2 details the proposed method using basic theories, including 4M theory, SPA, and SPA-based UAHP. A case study of an on-site fracturing operator is selected to illustrate the method in Section 3, where a comparison to the Fuzzy Comprehensive Evaluation Method (FCEM) is also provided. Finally, conclusions are made in Section 4.

Section snippets

Methodologies

There are many human factors that can lead to errors, incidents, and accidents. To assess the human error likelihood, an integrated method is proposed based on a novel UAHP-SPA model (see Fig. 2). It consists of a set of comprehensive assessment indexes, the UAHP weight optimization model, and the uncertainty-based assessment SPA technique.

Background

Shale-gas wells are located in mountainous areas, where fracturing operators face harsh conditions and physically-taxing operations. Consequently, there are serious human errors due to the psychological and physical fatigue. A case is described as follows: the pump pressure sharply declined at the last stage of hydraulic fracturing around 4:00 am due to a fatigued pump operator forgetting to close the butterfly valve. As a result, air came into the plunger and the pump failed.

In detail, the

Conclusions

Shale-gas fracturing requires an effective method to reduce human errors and ensure its safe operation. There have been many applications of human error likelihood assessment in oil and gas industry. However, the existing methods inadequately address the uncertainties from latent, unpredictable, and various human errors in fracturing operations. Using the UAHP-SPA model, an integrated assessment method has been proposed to improve on the existing studies. The assessment procedures have been

Acknowledgements

The authors greatly appreciate the guidance of Professor Carol Smidts and Dr. Yunfei Zhao during Qianlin’s academic visit at the Ohio State University. This work is supported by National Key Research and Development Program of China (2017YFC0805801), National Natural Science Foundation of China (No.51574263), Beijing Nova Program (Z181100006218048), and China Scholarship Council([2017]3109).

References (66)

  • M. Amiri et al.

    Fuzzy probabilistic expert system for occupational hazard assessment in construction

    Safety Sci.

    (2017)
  • B. Cai et al.

    A dynamic Bayesian networks modeling of human factors on offshore blowouts

    J. Loss Prevent. Proc.

    (2013)
  • T. Chong et al.

    Application of set pair analysis method on occupational hazard of coal mining

    Safety Sci.

    (2017)
  • S.A. Clancy et al.

    The potential for spills and leaks of contaminated liquids from shale gas developments

    Sci. Total Environ.

    (2018)
  • T. Dalijono et al.

    Reducing human error by improvement of design and organization

    Process Saf. Environ.

    (2006)
  • T. Deacon et al.

    Human error risk analysis in offshore emergencies

    Safety Sci.

    (2010)
  • T. Deacon et al.

    A framework for human error analysis of offshore evacuations

    Safety Sci.

    (2013)
  • D.G. DiMattia et al.

    Determination of human error probabilities for offshore platform musters

    J. Loss Prevent. Proc.

    (2005)
  • T. Engelder et al.

    Analysis of a gas explosion in Dimock PA (USA) during fracking operations in the Marcellus gas shale

    Process Saf. Environ.

    (2018)
  • R. Islam et al.

    Development of a monograph for human error likelihood assessment in marine operations

    Safety Sci.

    (2017)
  • R. Islam et al.

    Development of a human reliability assessment technique for the maintenance procedures of marine and offshore operations

    J. Loss Prevent. Proc.

    (2017)
  • R. Islam et al.

    Human error assessment during maintenance operations of marine systems – What are the effective environmental factors?

    Safety Sci.

    (2018)
  • R. Islam et al.

    Human error probability assessment during maintenance activities of marine systems

    Safety Health Work

    (2018)
  • F.I. Khan et al.

    HEPI: A new tool for human error probability calculation for offshore operation

    Safety Sci.

    (2006)
  • J.W. Kim et al.

    A taxonomy of performance influencing factors for human reliability analysis of emergency tasks

    J. Loss Prevent. Proc.

    (2003)
  • G. Landucci et al.

    A methodology for frequency tailorization dedicated to the Oil & Gas sector

    Process Saf. Environ.

    (2016)
  • C. Li et al.

    Risk assessment of water pollution sources based on an integrated k-means clustering and set pair analysis method in the region of Shiyan, China

    Sci. Total Environ.

    (2016)
  • M. Musharraf et al.

    Human reliability assessment during offshore emergency conditions

    Safety Sci.

    (2013)
  • A. Noroozi et al.

    Determination of human error probabilities in maintenance procedures of a pump

    Process Saf. Environ.

    (2014)
  • D. Rahm

    Regulating hydraulic fracturing in shale gas plays: The case of Texas

    Energ. Policy

    (2011)
  • G.O. Strand et al.

    Human factors modelling in offshore drilling operations

    J. Loss Prevent. Proc.

    (2016)
  • G.O. Strand et al.

    On the role of HMI in human reliability analysis of offshore drilling operations

    J. Loss Prevent. Proc.

    (2017)
  • S.C. Theophilus et al.

    Human factors analysis and classification system for the oil and gas industry (HFACS-OGI)

    Reliab. Eng, Syst. Safe.

    (2017)
  • Cited by (11)

    • Study on safety evaluation and design method of union joint thread for Ultra-high pressure fracturing manifold

      2022, Engineering Failure Analysis
      Citation Excerpt :

      Gao [13] et al. proposed a multi-axial random vibration fatigue damage parameter based on a critical plane damage parameter for multi-axial high cycle fatigue, which is an equivalent stress Power Spectral Density (PSD). In terms of safety assessment, so far, most of the studies have focused on the risk impact factor identification, risk scenario analysis and dynamic risk modeling of various pipelines [14–17]. There are few studies on failure evaluation of high pressure manifold [18,19], and a large amount of historical data and real-time detection technology are needed to support it.

    • Human Error Probability Determination in Blasting Process of Ore Mine Using a Hybrid of HEART and Best-Worst Methods

      2022, Safety and Health at Work
      Citation Excerpt :

      Human Reliability Analysis (HRA) [9] is the primary approach to prevent and reduce human error [9]. The HRA has three main steps: recognizing fundamental operations, analyzing relevant tasks, and determining human error probability (HEP) [10]. Several methods have been used to assess the human contribution to the accident, including Technique for Human Error Rate Prediction (THERP) [11], Simplified Plant Analysis Risk Human Reliability Assessment (SPAR-H) [12], Cognitive Reliability and Error Analysis Method (CREAM) [13], Information, Decision and Action in Crew context (IDAC) [14], Accident Sequence Evaluation Program (ASEP) [15], and Phoenix [16].

    • Simulation-based training to enhance process safety in offshore energy operations: Process tracing through eye-tracking

      2020, Process Safety and Environmental Protection
      Citation Excerpt :

      Given the broad spectrum of HFs and the complexity of offshore drilling activities, it comes as a challenge to observe, quantify, and optimize the HFs in drilling operations. Multiple techniques have been adopted to estimate and model risks involved in the drilling process (Abimbola and Khan, 2016; Abimbola et al., 2014; Wang et al., 2019). One way to look at HFs is through process-tracing techniques, where an attempt can be made to understand the dominant stimuli behind a certain decision or action.

    • Regression-based sparse polynomial chaos for uncertainty quantification of subsurface flow models

      2019, Journal of Computational Physics
      Citation Excerpt :

      Uncertainty Quantification (UQ) and Uncertainty Propagation (UP) in the subsurface flow related problems have been the subject of intensive research activities over the last decades [2–6].

    View all citing articles on Scopus
    View full text