Qualitative human error analysis

Information on the types of human interactions with hazardous systems that occur would be obtained from sources such as plant operating instructions, job safety analyses and similar sources. These interactions are referred to as critical tasks (CT). 

Given that workers interact with hazardous systems, how frequently are they likely to make errors in these critical tasks  

The answer to this question will depend on two factors the frequency with which the CT occur, and the likelihood of errors arising when performing these tasks. The frequency of the interactions can usually be specified relatively easily by reference to plant procedures, production plans, and maintenance schedules. The probability of error will be a fimction of the PIFs discussed extensively in Chapter 3 and other chapters in this book. In order to obtain a measure of error potential, it is necessary to make an assessment of the most important PIFs for each of the CT. 

In summary, at the screening stage of the SPEAR process, the ranking of tasks in order of potential risk is made on the basis of three criteria  

If these fimctions are each rated from 1 to 5, a scale of task criticallity can be generated ranging from 0 to 1 as follows  

---

The error reduction analysis concludes one complete cycle of the qualitative human error analysis component of the methodology set out in Figure 5.4. The analyst then decides if it is appropriate to perform a more detailed analysis on any of the operations considered at the current level. As a result of this process, operations 3.2 Monitor tanker following operation, 4.1 Stop filling operation, 4.2 Disconnect tanker, and 4.4 Secure tanker are analyzed in more detail (see Figure 5.6). 

The qualitative human error analysis stages described above are applied to the task steps in subtask 3.2. Examples of the results of this analysis are shown in Figure 5.8. The corresponding error-reduction strategies are shown in Figure 5.9. 

Qualitative human error prediction is the most important aspect of assessing and reducing the human contribution to risk. For this reason, it will be described in some detail in this section. The qualitative analysis performed in SPEAR involves the following techniques ... 

Several qualitative approaches can be used to identify hazardous reaction scenarios, including process hazard analysis, checklists, chemical interaction matrices, and an experience-based review. CCPS (1995a p. 176) describes nine hazard evaluation procedures that can be used to identify hazardous reaction scenarios-checklists, Dow fire and explosion indices, preliminary hazard analysis, what-if analysis, failure modes and effects analysis (FMEA), HAZOP study, fault tree analysis, human error analysis, and quantitative risk analysis. 

Historically, the emphasis in Human Reliability Analysis (HRA) has been on techniques for the derivation of Human Error Probabilities (HEPs) for use in systems analysis techniques such as FTA. However, HEA should be an integrated process that includes a systematic and rigorous qualitative analysis to identify the nature of the errors that can arise prior to any attempt at quantification. This qualitative Human Error Identification (HEI) must ensure that no significant failures are omitted from the analysis. 

Frequency Phase 1 Perform Qualitative Study, Typically Using HAZOP, FMEA, or What-if Analysis. To perform a qualitative study you should first (1) define the consequences of interest, (2) identify the initiating events and accident scenarios that could lead to the consequences of interest, and (3) identify the equipment failure modes and human errors that could contribute to the accident... 

A critical assembly is a split bed on which fissionable material used to mock up up a separated reactor core that is stacked half on each half. One half is on roller guides so that the two halves may be quickly pulled apart if the neutron multiplication gets too high. Use the Preliminary Hazards Analysis method described in section 3,2.1 to identify the possible accidents that may occur and the qualitative probabilities and consequences. List the initiators in a matrix to systematically investigate the whole process. Don t forget human error. 

When performing human reliability assessment in CPQRA, a qualitative analysis to specify the various ways in which human error can occur in the situation of interest is necessary as the first stage of the procedure. A comprehensive and systematic method is essential for this. If, for example, an error with critical consequences for the system is not identified, then the analysis may produce a spurious impression that the level of risk is acceptably low. Errors with less serious consequences, but with greater likelihood of occurrence, may also not be considered if the modeling approach is inadequate. In the usual approach to human reliability assessment, there is little assistance for the analyst with regard to searching for potential errors. Often, only omissions of actions in proceduralized task steps are considered. 

The use of a model of human error allows a systematic approach to be adopted to the prediction of human failures in CPI operations. Although there are difficulties associated with predicting the precise forms of mistakes, as opposed to slips, the cognitive approach provides a framework which can be used as part of a comprehensive qualitative assessment of failure modes. This can be used during design to eliminate potential error inducing conditions. It also has applications in the context of CPQRA methods, where a comprehensive qualitative analysis is an essential precursor of quantification. The links between these approaches and CPQRA will be discussed in Chapter 5. 

In addition, the chapter will provide an overview of htunan reliability quantification techniques, and the relationship between these techniques and qualitative modeling. The chapter will also describe how human reliability is integrated into chemical process quantitative risk assessment (CPQRA). Both qualitative and quantitative techniques will be integrated within a framework called SPEAR (System for Predictive Error Analysis and Reduction). 

If the results of the qualitative analysis are to be used as a starting-point for quantification, they need to be represented in an appropriate form. The form of representation can be a fault tree, as shown in Figure 5.2, or an event tree (see Bellamy et al., 1986). The event tree has traditionally been used to model simple tasks at the level of individual task steps, for example in the THERP (Technique for Human Error Rate Prediction) method for human reliability... 

PROBLEM DEFINITION. This is achieved through plant visits and discussions with risk analysts. In the usual application of THERP, the scenarios of interest are defined by the hardware orientated risk analyst, who would specify critical tasks (such as performing emergency actions) in scenarios such as major fires or gas releases. Thus, the analysis is usually driven by the needs of the hardware assessment to consider specific human errors in predefined, potentially high-risk scenarios. This is in contrast to the qualitative error prediction methodology described in Section 5.5, where all interactions by the operator with critical systems are considered from the point of view of their risk potential. 

There are many methods of safety analysis reviews that are available and can be applied to a facility or project design to overcome human errors and the various failures of the process system. The methods may be either qualitative or quantitative in nature. 

There are various types of analyses that are used for a process hazard analysis (PHA) of the equipment design and test procedures, including the effects of human error. Qualitative methods include checklists, What-If, and Hazard and Operability (HAZOP) studies. Quantitative methods include Event Trees, Fault Trees, and Failure Modes and Effect Analysis (FMEA). All of these methods require rigorous documentation and implementation to ensure that all potential safety problems are identified and the associated recommendations are addressed. The review should also consider what personal protective equipment (PPE) is needed to protect workers from injuries. 

Accuracy and Precision. The accuracy and precision required of an atomic-emission spectroscopic method affect the approach used in the analysis as well as the time involved. A qualitative analysis requires a minimum of effort, but as better accuracy and precision are demanded, increasing care is needed. Even if a representative sample has been obtained, errors inherent in the method, human errors, and random errors contribute to inaccuracies. Spectrochemical equipment is largely responsible for the random errors that influence precision, and both method and individual laboratory errors influence the accuracy. In addition, relative precision and accuracy depend upon concentration levels. The standard deviation increases with increasing concentration, but the relative standard deviation decreases the latter may vary from a few percent to less than one percent using photographic detection, depending on the element and the concentration. 

Analytical methodology was developed for accurate quantitative analysis of trichothecenes at low part-per-billion levels in blood. Although this methodology was arduous and lacked the ruggedness normally demanded of an analytical procedure which must nave a low failure rates it proved to be both qualitatively reliable and quantitatively accurate when it was combined with a well planned quality assurance program. An indispensable part of developing the quality assurance plan was a formal risk assessment which specifically took into account the possibility of human error. 

It is fundamental for assessing human error in systems analyses to identify and describe the human acts with importance for the event sequence under analysis (qualitative assessment). This corresponds to the task analyses, which are characteristic of ergonomic studies. Firstly, the important actions, the moment in time at which they are required and the time period available for their execution have to be determined. Furthermore, the requirements for the action, the information necessary, respectively available, the possibilities of correction in case of omission or faulty execution must be estabhshed. Additionally, other factors of important influence on human reliabihty such as the state of knowledge on the process in question, ergonomically favourable or disadvantageous layout of the workplace, the tools or the environment are identihed. On the basis of this task analysis reliability data (normally failure probabilities on demand) are assigned to the tasks identified. They stem from existing data collections (cf. Table 9.21). 

Fault tree analysis is a technique by which the system safety engineer can rigorously evaluate specific hazardous events. It is a type of logic tree that is developed by deductive logic from a top undesired event to all subevents that must occur to cause it. It is primarily used as a qualitative technique for studying hazardous events in systems, subsystems, components, or operations involving command paths. It can also be used for quantitatively evaluating the probability of the top event and all subevent occurrences when sufficient and accurate data are available. Quantitative analyses shall be performed only when it is reasonably certain that the data for part/component failures and human errors for the operational environment exist. 

Fault Tree Analysis (FTA) is a well known and widely used safety tool, implementing a deductive, top down approach. It starts with a top level hazard, which has to be known in advance and "works the way down" through all causal factors of this hazard, combined with Boolean Logic (mainly AND and OR gates). It can consider hardware, software and human errors and identifies both single and multiple points of failure. Both a quantitative and qualitative analysis is possible. 

Human reliability analysis (HRA) With technological development and incorporation of redundancy it is possible to reduce equipment failure to a great extent. However, human behavior is not that predictable. So, there are chances that failure could occur because of human factors. This is a method by which probability is measured. It is also used in PFLA. This could be quantitative as well as qualitative. Although the exact value is not certain it is estimated that error committed by a human could be as high as 60—80% (even 90%). Human performance is affected by several factors, referred to as the performance shaping factor (PSF). By this method, PSF is identified and tries to improve it. In addition to PSF, normal human error probability (HEP) is also calculated on the basis of human activity. There are so many factors that affect this analysis accuracy, reproducibility, bias, etc. There have been several methods and each needs to be understood before application. An HRA event tree is often used. It may be informative to refer to Table V/1.0-1 (Chapter V). 

The objective of human factors safety analysis is to identify and correct human error situations that could lead to significant hazards. The analysis can be either qualitative or quantitative, depending on the level of detail desired and what the consequences are of a person making a mistake. The steps of a human factors safety analysis are as follows ... 

Traditional methods of additive analysis and the required instruments are often expensive and require the efforts of a skilled technician or chemist. In some cases a single instmment can not provide analyses for the wide variety of additives a particular organisation utilises. Additionally, laboratory techniques rarely provide results in a timely fashion. Determination of physical properties is not the least important if one thinks of pigments, talc and other fillers. Application of spectroscopic techniques to polymer production processes permits real-time measurement of those qualitative variables that form the polymer manufacturing specification, i.e. both chemical properties (composition, additive concentration) and physical properties (such as melt index, density). On-line analysis may intercept plant problems such as computer error, mechanical problems and human error with respect to additive incorporation in the resin production. Characterisation and quantitative determination of additives in technical polymers is an important but difficult issue in process and quality control. 

Several methods to quantify human error probability have been reviewed in Section 9.2. These methods suffer from the difficulty associated with any attempt to construct quantitative, predictive models of human behaviour. The qualitative methods on the other hand, require multi-disciplinary teams to carry out an analysis and this is regarded as being resource intensive. The more recent HRA methods have included cognitive aspects of decision making and the time dimension. However, it has not yet captured the fundamental nature of the interaction between actions and machine responses (Cacciabue et al. (1993)). These interactions lie in the mutual dynamic influence of the operator, the plant and the interfaces. 

The goal of any statistical analysis is inference concerning whether on the basis of available data, some hypothesis about the natural world is true. The hypothesis may consist of the value of some parameter or parameters, such as a physical constant or the exact proportion of an allelic variant in a human population, or the hypothesis may be a qualitative statement, such as This protein adopts an a/p barrel fold or I am currently in Philadelphia. The parameters or hypothesis can be unobservable or as yet unobserved. How the data arise from the parameters is called the model for the system under study and may include estimates of experimental error as well as our best understanding of the physical process of the system. 

Performance-influencing factors analysis is an important part of the human reliability aspects of risk assessment. It can be applied in two areas. The first of these is the qualitative prediction of possible errors that could have a major impact on plant or personnel safety. The second is the evaluation of the operational conditions under which tasks are performed. These conditions will have a major impact in determining the probability that a particular error will be committed, and hence need to be systematically assessed as part of the quantification process. This application of PIFs will be described in Chapters 4 and 5. 

Because most research effort in the human reliability domain has focused on the quantification of error probabilities, a large number of techniques exist. However, a relatively small number of these techniques have actually been applied in practical risk assessments, and even fewer have been used in the CPI. For this reason, in this section only three techniques will be described in detail. More extensive reviews are available from other sources (e.g., Kirwan et al., 1988 Kirwan, 1990 Meister, 1984). Following a brief description of each technique, a case study will be provided to illustrate the application of the technique in practice. As emphasized in the early part of this chapter, quantification has to be preceded by a rigorous qualitative analysis in order to ensure that all errors with significant consequences are identified. If the qualitative analysis is incomplete, then quanhfication will be inaccurate. It is also important to be aware of the limitations of the accuracy of the data generally available... 

Lastly, the quality of a statistical correlation alone cannot be taken as an indication of correctness of the assumptions. For example, a model with slightly larger, but random error is more likely correct than its rival with smaller, but systematic error, and primitive statistics programs do not take this into account. As Connors puts it, "the human eye, in combination with chemical knowledge, is a more subtle qualitative judge of data than is regression analysis" [57]. 

Most HR A methods use PSFs to support the analysis of the contextual conditions under which human failure may occur. In the qualitative HRA, PSFs help analysts to systematically address all factors believed to be important to human performance. In the quantitative HRA, PFSs are frequently used to modify a generic error probability to capture the effect of the context in which the task of interest is carried out. 

---