NUREG/CR-6883

INL/EXT-05-00509

The SPAR-H Human

Reliability Analysis Method

Idaho National Laboratory

U.S. Nuclear Regulatory Commission

Office of Nuclear Regulatory Research
Washington, DC 20555-0001
 NUREG/CR-6883
INL/EXT-05-00509

The SPAR-H Human

Reliability Analysis Method

Manuscript Completed: September 2004

Published: August 2005

Prepared by
D. Gertman, H. Blackman,
J. Marble, J. Byers, C. Smith

Idaho National Laboratory

Battelle Energy Alliance
Idaho Falls, ID 83415

P. O’Reilly, NRC Project Officer

Prepared for
Division of Risk Analysis and Applications
Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission
Washington, DC 20555-0001
NRC Job Code W6355
 ABSTRACT
In support of the Accident Sequence Precursor by staff analysts, and from other experience gained
Program (ASP), the U.S. Nuclear Regulatory at the INL in applying the method in human
Commission (NRC), in conjunction with the Idaho reliability analysis (HRA), a number of needed
National Laboratory (INL), in 1994 developed the improvements to definitions, terms, and concepts
Accident Sequence Precursor Standardized Plant were identified. In 2003, to enhance the general
Analysis Risk Model (ASP/SPAR) human utility of the SPAR-H method and to make it more
reliability analysis (HRA) method, which was used widely available, the method was updated and
in the development of nuclear power plant (NPP) reviewed for its applicability to low-power and
models. Based on experience gained in field- shutdown applications. During this review, an
testing, this method was updated in 1999 and approach to uncertainty representation was
renamed SPAR-H, for Standardized Plant Analysis outlined, based on the beta distribution. Additional
Risk-Human Reliability Analysis method. Since detail regarding human error probability (HEP)
that time, NRC staff analysts have been using this dependency assignment was also made available.
method to perform their risk-informed regulatory
activities, such as determining the risk significance This document presents the current version of the
of inspection findings in Phase 3 of the SPAR-H method, along with guidance, definitions,
Significance Determination Process, developing an improvements in representing uncertainty, and
integrated risk-informed performance measure in increased detail regarding dependency assignment
support of the reactor oversight process, and for HEP calculations. This report also contains
systematically screening and analyzing operating comparisons between this and other contemporary
experience data in order to identify HRA approaches and findings specific to
events/conditions that are precursors to severe application of the method to low power and
accident sequences. As a result of implementation shutdown events.

iii
iv
 FOREWORD
In the early 1990s, the U.S. Nuclear Regulatory Commission (NRC) identified the need for an
improved, traceable, easy-to-use human reliability analysis (HRA) method for use with the
analytical models associated with the agency’s Accident Sequence Precursor (ASP) Program.
This report documents the most recent update of the Standardized Plant Analysis Risk (SPAR) HRA
(SPAR-H) Method, which evolved in response to this need.
Initially, the NRC contracted with Idaho National Laboratory (INL) 1 to develop the “ASP SPAR
HRA Method,” which consisted of a two-step process to identify nominal human error
probabilities (HEPs), and then modify those HEPs on the basis of summary-level performance-
shaping factors (PSFs) and dependence. Significantly, this method required analysts to complete
a relatively straightforward worksheet, which was then used to estimate the PSFs and the HEP of
interest. Then, in 1999, the NRC directed INL to update the ASP SPAR HRA Method by modifying
the PSFs, dependencies, and base HEPs using a benchmarking process, and the modified method
was renamed as the “SPAR-H Method.” Most recently, in 2002–2003, the NRC asked INL to
update the model to (1) improve definitions, terms, and concepts; (2) produce a reference
document; (3) review the applicability of the SPAR-H Method to low-power and shutdown
applications; (4) develop a treatment approach for uncertainties associated with human performance
parameters; and (5) present additional detail regarding assignment of HEP dependencies. This report
presents the results of this work, which have undergone peer review by internal and external
stakeholders.
Over time, NRC analysts have come to use the SPAR models and SPAR-H Method extensively in
performing their risk-informed regulatory activities in a variety of agency programs. The SPAR-H
Method is an adequate HRA tool for use with the SPAR models in performing risk analyses of
operational events/conditions. In particular, the affected programs include the ASP Program, the
Significance Determination Process (SDP), generic issue resolution, and license amendment
reviews. Nonetheless, as a simplified method, SPAR-H has inherent modeling and analysis
limitations that should be clearly understood. The SPAR-H Method should not necessarily be
preferred over more sophisticated and detailed approaches, such as A Technique for Human
Event Analysis (ATHEANA), in situations that require detailed analysis of the human
performance aspects of an event.

Carl J. Paperiello, Director

Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission

1
Idaho National Laboratory was formerly known as Idaho National Engineering and Environmental Laboratory (INEEL).
The name change occurred in February 2005, when the U.S. Department of Energy entered into an agreement with a new
contractor to manage the laboratory.
v
vi
 CONTENTS
ABSTRACT.................................................................................................................................................iii

FOREWORD ................................................................................................................................................ v

EXECUTIVE SUMMARY .......................................................................................................................xiii

ACKNOWLEDGEMENTS .....................................................................................................................xxiii

ACRONYMS............................................................................................................................................ xxv

GLOSSARY ........................................................................................................................................... xxvii

1. INTRODUCTION.............................................................................................................................. 1

1.1 Overview.................................................................................................................................. 1

1.2 Background .............................................................................................................................. 1

1.3 HRA Orientation ...................................................................................................................... 2

1.3.1 Guidance in performing HRA ....................................................................................... 2

1.4 Organization............................................................................................................................. 3

2. SPAR-H METHOD............................................................................................................................ 5

2.1 Model of Human Performance................................................................................................. 5

2.1.1 The Role of Work Processes.......................................................................................... 9

2.2 Task Types ............................................................................................................................. 10

2.2.1 Guidance for Diagnosis ............................................................................................... 10

2.2.2 Guidance for Action .................................................................................................... 10
2.2.3 Guidance for Diagnosis and Action............................................................................. 10
2.3 Error Types ............................................................................................................................ 11

2.4 PSFs ....................................................................................................................................... 12

2.4.1 PSF Comparison Findings ........................................................................................... 17

2.4.2 PSF Changes................................................................................................................ 18
2.4.3 Relationship of PSFs to HEPs Underlying the SPAR-H Method................................ 18
2.4.4 SPAR-H Method PSF Overview and Definitions........................................................ 20
2.5 Application of Multiple PSFs................................................................................................. 27

2.5.1 Calculating the Composite PSF ................................................................................... 28

2.6 Dependency............................................................................................................................ 29

2.7 Uncertainty Analysis Suggestions For Using SPAR-H ......................................................... 31

2.7.1 Overview ..................................................................................................................... 31

vii
 2.7.2 Human Performance Distributions .............................................................................. 32
2.7.3 Work Shift Effects ....................................................................................................... 35
2.7.4 Human Performance and Complexity ......................................................................... 36
2.7.5 The Categorization and Orthogonality of PSFs........................................................... 38
2.7.6 The CNI Distribution................................................................................................... 39
2.7.7 Combining Non-SPAR-H Information with SPAR-H................................................. 40
2.8 Recovery ................................................................................................................................ 42

3. ANALYSIS ...................................................................................................................................... 43

3.1 Base Rate Comparison Among HRA Methods, Including SPAR-H ..................................... 43

3.2 Comparison of PSF Weights for Low Power Versus At-power ............................................ 46

3.3 Approach to LP/SD Comparison ........................................................................................... 48

3.4 Additional Field Testing ........................................................................................................ 49

3.4.1 Applicability of the SPAR-H Method to External Events........................................... 49

3.5 Range of Uncertainty Associated with HRA Methods .......................................................... 50

3.5.1 Evaluation Against Other Methods ............................................................................. 50

3.5.2 Change of Distribution Due to Truncation .................................................................. 50
3.6 Change in Time PSF .............................................................................................................. 50

4. USING SPAR-H............................................................................................................................... 55

4.1 Modeling Conventions and Considerations ........................................................................... 55

4.2 At-power ................................................................................................................................ 56

4.2.1 Prerequisites................................................................................................................. 56
4.2.2 ATHEANA Search Process......................................................................................... 56
4.2.3 Using the SPAR-H Method for a SPAR Base Model.................................................. 57
4.2.4 Using the SPAR-H Method for SPAR Event Analysis ............................................... 58
4.2.5 Sources of Information for Applying the SPAR-H Method to Events ........................ 58
4.2.6 Completing the SPAR-H Human Error Worksheet..................................................... 59
5. DISCUSSION................................................................................................................................... 65

5.1 Differences between At-power and LP/SD............................................................................ 65

5.2 Compliance with ASME Standard on PRA ........................................................................... 66

5.2.1 Organization ................................................................................................................ 66

5.2.2 Documentation............................................................................................................. 66
5.2.3 Expert Judgment .......................................................................................................... 67
5.2.4 Activity Types ............................................................................................................. 67
5.2.5 Work Processes............................................................................................................ 67
5.2.6 Probability Assignment ............................................................................................... 67
5.2.7 PSF Inclusion............................................................................................................... 67
5.2.8 Dependency and Procedures........................................................................................ 68

viii
 5.2.9 Procedures Review and Documentation ...................................................................... 68
5.2.10 Supporting Requirements for HRA ............................................................................. 68
5.2.11 Recovery...................................................................................................................... 68
5.2.12 Timing ......................................................................................................................... 69
5.2.13 Screening ..................................................................................................................... 69
5.2.14 Task Characteristics..................................................................................................... 69
5.3 NASA Guidelines .................................................................................................................. 69

5.4 General Discussion ................................................................................................................ 70

6. REFERENCES ................................................................................................................................. 75

ix
x
 Appendices

Appendix A—HRA Worksheets for At-power......................................................................................... A-1

Appendix B—HRA Worksheets for LP/SD ............................................................................................. B-1

Appendix C—Full Power Worksheets for SGTR Example...................................................................... C-1

Appendix D—LP/SD Worksheets for PWR LOI with RCS Pressurized ................................................. D-1

Appendix E—Worksheets for Dry Cask................................................................................................... E-1

Appendix F—Operational Examples of SPAR-H Method Assignement of PSF Levels...........................F-1

Appendix G—The Relative Relationship Among SPAR PSFs ................................................................ G-1

Appendix H—SPAR Development History ............................................................................................. H-1

Appendix I—SPAR-H Review Comments.................................................................................................I-1

Figures
Figure ES-1. Ideal mean HEP as a function of the influence of performance shaping factors.................. xiv

Figure 2-2. Ideal mean HEP as a function of PSF influence. ..................................................................... 19

Figure 2-3. Factors contributing to task complexity. ................................................................................. 22

Figure 2-4. Arousal effect on memory........................................................................................................ 34

Figure 2-6. Alpha (α) as a function of mean HEP. ..................................................................................... 41

Figure 2-7. CNI distribution for the HEP. .................................................................................................. 41

Figure 4-1. Basic flow diagram for completing the SPAR-H worksheets. ................................................. 61

Figure D-1. Loss of inventory event tree with RCS pressurized for a nuclear power plant. ...................D-4

xi
 Tables
Table 2-1. Operational Factors in SPAR-H .................................................................................................. 8

Table 2-2. HRA methods used in SPAR-H comparisons. .......................................................................... 13

Table 2-3. Action PSF Comparison Matrix, at power (PSFs = 8). ............................................................. 14

Table 2-4. SPAR-H Dependency Rating System........................................................................................ 30

Table 3-1. Action error type base rate comparison. .................................................................................... 43

Table 3-2. Mixed-task base rate comparison. ............................................................................................. 44

Table 3-3. Diagnosis error type base rate comparison................................................................................ 44

Table 3-4. SPAR-H PSFs used in quantifying HEPs.................................................................................. 45

Table 3-5 Assumed differences among LP/SD conditions and at-power mode. ........................................ 47

Table 3-6. Loss of inventory with RCS pressurized HEPs Comparison of PSF influence for PSF Weight

Sets A and B........................................................................................................................................ 49

Table 3-7. Diagnosis and action error factors as a function of HRA method. ............................................ 51

Table 3-8. Available time PSF influence for LP/SD.................................................................................. 52

Table 3-9. Influence of expansive time on base failure rates...................................................................... 53

Table 4-1. PSF sources of information for SPAR-H................................................................................... 59

Table 5-1 SPAR-H method assessment. ..................................................................................................... 72

Table G-1. The relative relationship among SPAR-H PSFs. ....................................................................G-3

Table I-1. Formal peer review comments and responses.......................................................................... I-4

xii
 EXECUTIVE SUMMARY
Human performance has been a key component of Based on review of first- and second-generation
incidents and accidents in many industries. HRA methods, the SPAR-H method assigns
Recently, the role of human error was documented human activity to one of two general task
in a number of well studied, high-profile events in categories: action or diagnosis. Examples of action
the nuclear power industry (Gertman et al. 2002). tasks include operating equipment, performing
Studies of these events included human reliability line-ups, starting pumps, conducting calibration or
analysis (HRA). Human reliability analysis is an testing, and other activities performed during the
evolving field that addresses the need to account course of following plant procedures or work
for human errors when: (a) performing safety orders. Diagnosis tasks consist of reliance on
studies such as probabilistic risk analysis (PRA); knowledge and experience to understand existing
(b) helping to risk-inform the inspection process; conditions, planning and prioritizing activities, and
(c) reviewing special issues; and (d) helping to determining appropriate courses of action. Base
risk-inform regulation. HRA has also been used to error rates for the two task types associated with
support the development of plant-specific PRA the SPAR-H method were calibrated against other
models. HRA methods. The calibration revealed that the
SPAR-H human error rates fall within the range of
This report presents a simple HRA method for rates predicted by other HRA methods.
estimating the human error probabilities associated
with operator and crew actions and decisions in A number of HRA methods do not have an explicit
response to initiating events at commercial U.S. human performance model. The SPAR-H method
nuclear power plants (NPPs). The Standardized is built on an explicit information-processing
Plant Analysis Risk Human Reliability Analysis model of human performance derived from the
(SPAR-H) method was developed to support behavioral sciences literature that was then
development of plant-specific PRA models for the interpreted in light of activities at NPPs (Blackman
U.S. Nuclear Regulatory Commission (NRC), and Byers 1994). In 1999, further research
Office of Regulatory Research (RES), and recently identified eight PSFs capable of influencing
has been used to help support the Office of human performance. These PSFs are accounted for
Reactor Regulation (NRR) Reactor Oversight in the SPAR-H quantification process. These
Process (ROP). The SPAR-H method is also factors include:
applicable to preinitiator events.
• Available time
The basic SPAR-H framework: • Stress and stressors
• Decomposes probability into contributions • Experience and training
from diagnosis failures and action failures • Complexity
• Accounts for the context associated with • Ergonomics (including the human-machine
human failure events (HFEs) by using interface)
performance-shaping factors (PSFs), and
dependency assignment to adjust a base-case • Procedures
HEP • Fitness for duty
• Uses pre-defined base-case HEPs and PSFs, • Work processes.
together with guidance on how to assign the
appropriate value of the PSF While many contemporary methods address PSFs
in some form, the SPAR-H method is one of the
• Employs a beta distribution for uncertainty few that addresses the potential beneficial
analysis influence of these factors. That is, positive
• Uses designated worksheets to ensure analyst influences of PSFs can operate in some instances
consistency. to reduce nominal failure rates. For example,

xiii
superior experience and training can serve to subsequent failures; in these situations, analysts
enhance the operator’s understanding of system are expected to use nominal rates when
status beyond the average or nominal case. This determining the HEP for subsequent failures.
does not mean that the operator or crew’s
knowledge is necessarily complete, merely that it Although the literature on dependency among
is better by some objective measure, which can human errors is limited, the INL review
enhance performance. Figure ES-1 shows this determined that the presence of the following
relationship and the influence of the PSF (x-axis) combinations of factors contributes to error
on mean human error probability (HEP) values dependency:
(y-axis).
• Same crew (relates to similar mindset, use of
The SPAR-H method addresses dependency. similar heuristics, tendencies to tunnel vision,
Dependency, in this case, means that the negative recency effects, etc.)
influence of a human error on subsequent errors is
• Same location (the control, display, or
accounted for by the model and is reflected in
equipment must be the same or located within
calculating the HEP. The model does not explicitly
the same relatively restricted area, such as the
address the influence of positive dependency on
same panel)

Greater human error

probability
1.0

Stronger error
causing effect
of the PSF
Stronger performance
enhancing effect Nominal error rate
of the PSF (1.0 E-2 for diagnosis,
1.0E-3 for actions

Lower human error

probability
1E-5

Figure ES-1. Ideal mean HEP as a function of the influence of performance shaping factors.

xiv
• Lack of additional cues [additional cues exist the results of each, and then select the most
if there is a specific procedural callout or a appropriate decomposition level.
different procedure is used, or additional
alarm(s) or display(s) are present] SPAR-H has been used in the development of
plant models and in event analysis, and it is
• Close succession of the next HEP (from possible to apply the method to retrospective as
within seconds to a few minutes). well as prospective scenarios. The criterion for
applying the SPAR-H method dependency
Various combinations of these contributory factors
assignment is the same for either case.
were examined and given a rating based on their
combined effect on dependency among tasks. The The application of the SPAR-H method is
ratings of the various combinations correspond to relatively straightforward and follows the guidance
zero, low, moderate, high, or complete for conducting HRA, which is available in a
dependency among tasks. In integrating this number of publicly available sources. Such
dependency information, the SPAR-H method sources include IEEE Standard 1082 for HRA
uses the underlying THERP quantification (1997), ASME Standard for Probabilistic Risk
provided in NUREG/CR-1278 (1983), but offers Assessment (ASME STD-RA-S-2002), and
an improved basis for dependency assignment. EPRI’s 1984 Systematic Human Action Reliability
Procedure (SHARP; Hannaman and Spurgin
Once dependency has been determined to be
1984). A number of analysts within the industry
present, moderate-to-high dependency will
may also have access to SHARP1 (Wakefield,
dominate the failure rate obtained when applying
Parry, and Spurgin 1990), but distribution is
the SPAR-H method; however, satisfying the
limited. When applied to situations other than
requirements for this level of dependency is not
SPAR model building or screening situations, the
often met. This restriction occurs because many
comprehensive HRA search strategies found in
actions involve different steps in procedures and
NUREG-1624 (2000) can be used to aid in
provide for relatively long periods of time between
identifying and modeling errors leading to unsafe
actions. In addition, the location of the equipment
acts and human failure events.
acted upon is not similar. Conversely, dependency
assignment is almost always applicable in The SPAR-H method produces a simple best
situations where an HRA analyst is attempting to estimate for use in plant risk models. The mean is
model the influence of a second or third checker in assumed to be the best (i.e., most informative)
a recovery sequence. piece of information available regarding the
human error probability. In addressing uncertainty,
The SPAR-H method may be applied on a task error factors were not used, and the use of a
level [as is often the case when developing SPAR lognormal probability distribution was not
models for low power/shutdown (LP/SD) or at- assumed. The SPAR-H method employs a beta
power], or on a subtask level when building HRA distribution, which can mimic normal and
event trees or fault trees (i.e., performing more lognormal distributions, but it has the advantage
detailed analysis). Once a decision regarding the that probabilities calculated with this approach
level of decomposition has been determined, the range from 0 to 1. A constrained noninformative
application of SPAR-H on either the task or prior, based on Atwood (1996), was selected for
subtask level should be consistent within the PRA. its ability to preserve the overall mean value while
While minor differences in HEP estimates for producing values at the upper end of the
failure events may be exhibited on the basis of the distribution that more accurately represent the
level of decomposition selected, this problem expected error probability. Analyses contained in
should not dominate findings of the risk analysis this report also review human performance
and is not unique to SPAR-H. In the event that distributions, relate them to performance shaping
there are applications where the level of event factors, and discuss issues regarding the relative
decomposition is thought to lead to different orthogonality of performance shaping factors’
results, we suggest that the analyst perform the influence on human performance.
analysis at both levels of decomposition, review

xv
A major component of the SPAR-H method is the dependency is then used to support quantification
SPAR-H Worksheet, presented in Appendices A of the final HEP.
and B. The method for filling out these worksheets
is described in this report. Note that the process Since there is some overlap among PSFs, there is a
differs slightly, depending on whether the analyst possibility that an influence might be double
is using the method to build SPAR models, counted, and analysts should be cautious in this
perform event analysis, or perform a more detailed regard. In addition, in highly negative situations,
HRA analysis. The analysis presented below refers i.e., strong negative contexts, where the propensity
to the use of the SPAR-H method to support SPAR for error is high, it is possible that analysts’
PRA model development, the major focus for the assignment of PSF levels can result in the
HRA method development process. calculation of a mean that would be numerically
larger than 1. In previous versions of the SPAR-H
SPAR-H WORKSHEET PROCESS method, the general guidance was to round the
OVERVIEW HEP estimate to 1. A mathematical solution for
this problem was sought, and a corresponding
In most instances, the HRA analyst will review adjustment factor was developed to avoid
SPAR model event trees containing action or probability estimates greater than 1. This
diagnosis tasks and accompanying contextual adjustment factor for use of multiple negative
information for consideration and evaluation. In PSFs is presented on the worksheets. We suggest
the majority of instances, the event will require that the adjustment factor be used in situations
analysis on a task level, that is, multiple subtasks where at least three nonnominal (negative) PSFs
are considered. Event trees and a limited number have been identified. (For a more detailed review
of fault trees will be available from the PRA see Section 2.5.) No adjustment factor for positive
analyst. The HRA analyst will determine whether PSFs was developed. The positive influence of
actions specified involve diagnosis or are purely dependency has not been investigated and
action-based. In some instances, action and therefore is not part of the SPAR-H method. A
diagnosis are intertwined and indiscernible. In lower bound cut-off of 1.0E-5 for HEPs is
others, a step in SPAR events may represent a task suggested.
with many underlying subtasks, including
planning or diagnosis. In such instances, the basic SPAR-H WORKSHEET PROCESS
event in the PRA model represents both diagnosis The mechanics of completing the SPAR-H human
and action. If a task involves both action and error worksheets are as follows.
diagnosis, two worksheets corresponding to action
and diagnosis are filled out, and a joint HEP is Step 1. Enter header information. This
calculated. This event is later reviewed for information refers to the:
dependency (see below).
• Plant being rated
When developing the basic SPAR-H model, three
• Name of the initiating event [e.g., partial loss
of the eight PSFs are evaluated: time available,
of offsite power (LOSP)]
stress and stressors, and complexity. The
remaining five PSFs (experience, procedures, • Basic event code [e.g., failure to restore one
ergonomics and human-machine interface, fitness of the emergency diesel generators (EDG),
for duty, and work processes) are generally rated XHE – LOSP – EDG]
nominal, because they are usually event-, plant-, or
• Coder ID [i.e., name of the analyst filling out
personnel-specific. These five PSFs are evaluated
the worksheet(s)]
when a plant-specific model is being developed.
• Context of the basic event being rated (e.g.,
Following determination of task category, the previous events in this particular sequence)
relationship of a failed task to a preceding failed
task (i.e., the task dependency) is assessed • General description of the event being rated
according to SPAR-H definitions. This (e.g., operator fails to perform correct action).

xvi
Step 2. Decide whether the basic event involves LP/SD Worksheet Enhancements
diagnosis, action, or both diagnosis and action.
PSF Time Available for Actions. The dynamic
Step 3. If diagnosis is involved, then rate the eight range of influence for expansive time available
PSFs according to the guidance provided. Use one defined as 50x nominal, was changed. The range
check mark for each PSF. Note any time a of effect for expansive time now reduces the
nonnominal PSF value is selected; document the nominal rate by a multiplier of 0.01. When time
reason for both nominal and nonnominal PSFs. available is determined to be 5x nominal, then a
Step 4. Transfer the multipliers to the calculation multiplier of 0.1 is used.
portion of the worksheet.
PSF Procedures. An additional level of influence
Step 5. Determine the HEP without dependency was incorporated for LP/SD. Incomplete or partial
(Pw/od). If there are 3 or more negative PSFs, then procedures influence the base HEP by a factor of
apply the adjustment factor provided on the 20. This influence is present for action and
worksheets (Section 2.5). diagnosis tasks. Analysts commented that this
Step 6. If action is involved, repeat Steps 3–5 for assignment would potentially be valuable for at-
the action portion. power conditions as well, and this level has also
been applied to the at-power worksheets.
Step 7. Calculate the overall HEP total, using the
diagnosis HEP, the action HEP, or the joint HEP. PSF Time Available for Diagnosis. Time available
multipliers were developed. Diagnoses with extra
Step 8. Determine the appropriate level of
available time available ranging from between 1 to
dependency from the table. If there is no 2x nominal are assigned a multiplier of 0.1;
dependency, document why there is none in the
expansive time (defined as >2x nominal) is
space provided. If there is no dependency, then the
assigned a range of effect from 0.1 to 0.01, which
total HEP is that produced in Step 7.
may be assigned by the analyst. This better reflects
Step 9. If dependency criteria are met and a level the increased uncertainty and longer time horizons
of dependency has been assigned, then calculate associated with a number of LP/SD tasks.
the task failure probability with formal
dependence, by using the worksheet. PSF Complexity. An additional level of influence
for favorable complexity (i.e., obvious diagnosis)
Appendices C, D, and E present examples of the was developed. The multiplier associated for this
worksheet. category is 0.1.

COMPARISON OF LP/SD AND At-power Worksheet Enhancements

AT-POWER CONDITIONS PSF Time Available for Actions. The dynamic
As a result of a qualitative comparison of LP/SD range of influence for expansive time available
and at-power condition, we determined that a was changed. The range of effect for expansive
separate worksheet for LP/SD and at power time now uses a multiplier of 0.01. In addition to
conditions should be developed. The worksheet for the use of absolute minutes in earlier versions of
at power is presented in Appendix A, while the SPAR-H, the relative time available in conjunction
worksheet for LP/SD is presented in Appendix B. with the time required for task performance has
Findings from field test results of the SPAR-H been taken into account.
method led to improvements in the LP/SD and at- PSF Procedures. An additional level of influence
power worksheets. The following enhancements to was incorporated. Incomplete or partial procedures
earlier versions of the worksheet were influence the base HEP by a factor of 20. This
implemented: influence is present for action and diagnosis tasks.

xvii
PSF Time Available for Diagnosis. The influence If, in the judgment of the PRA and HRA analyst,
of expansive time (>24 hours for diagnosis for at- additional detailed analyses are called for, the
power conditions) was changed from a multiplier tendency for either omissions or commissions to
of 0.001 to 0.01. be more important in contributing to an individual
human failure event can be explicitly modeled by
An additional PSF category, “Insufficient the analyst. For example, the subtask level of
Information Available,” was added to both decomposition can be used when building
worksheets for each individual PSF. supporting fault trees.

DISCUSSION The explicit incorporation of work processes in

PRA/HRA is relatively new. First generation
The SPAR-H method is straightforward, easy to methods acknowledged work practices when
apply, and is based on a human information- taking into account the use of a second checker or
processing model of human performance and procedure quality. In instances where the work
results from human performance studies available process PSF is thought to influence performance,
in the behavioral sciences literature (Newell and it is often difficult to determine its effects. This is,
Simon 1972). This simplified HRA approach in part, because the effects of work processes and
contains a number of significant features, organizational factors are often diffuse. For
including calibration of its base failure rates and example, the amount of workarounds, failure to
range of PSFs’ influence with other HRA trend problems, and failure to respond to industry
methods. This version of the SPAR-H method also notices may increase the likelihood of equipment
contains a revised approach to uncertainty unavailability, increase the likelihood of errors,
analysis, employing a beta distribution that and/or reduce the likelihood of error recovery.
obviates problems experienced in earlier versions
when applying error factor approaches, and an The range of effect used in SPAR-H reflects the
adjustment factor for situations where the estimate treatment of the work process PSF in other HRA
of the mean HEP is greater than 1. methods. For example, work processes range of
effect in SPAR-H is enveloped by identification of
SPAR-H has been refined as a result of experience a range of effect for work process PSF in two
gained during its use in the development of over methods, CREAM (Hollnagel 1998) and HEART
70 SPAR PRA plant models for the NRC, in (Williams 1992). The work process PSF definition
limited HRA applications for dry cask spent-fuel owes some debt to the work process analysis
storage, in implementation of risk-informed plant method (WPAM) found in Weil and Apostolakis
inspection notebooks, and through third party (2002). Obviously, other HRA methods, including
application to other domains such as aerospace. those making use of simulator trials such as
The method does not differentiate between active MERMOS (Bieder et al. 1999), do not have a
and latent failures. Identification and modeling of direct work process parameter from which a range
human failure as either active or latent is the of effects could be used to inform SPAR-H. The
decision of the analyst. It is thought that the same same is true for the ORE/CBDT (EPRI TR-100259
PSFs and base failure rates are applicable to either 1992) approach. The ORE/CBDT provides
type of error. The base error rates contained in the normalized time reliability curves, a cue response
worksheets for actions and diagnosis include model, and yes/no decision trees for different
omission and commission types of errors; the failure modes. Formal communication and
explicit representation of omission versus compliance failures are represented in a subset of
commission is an issue left to the analyst and is the trees. Other aspects of work practices may be
part of the error identification and modeling implicitly present in the time reliability curves.
process constituting HRA. This is in contrast to However, extracting the relative range of effect is
other, more in-depth methods such as ATHEANA, somewhat difficult and the ORE/CBDT model was
which focuses on the identification and not designed with balance of plant operations or
quantification of errors of commission. preinitiator failures in mind. Thus, the range in

xviii
SPAR-H is within the bounds suggested by the An example of the type of information not covered
other HRA methods above. in great detail in this report is a tutorial on how to
construct fault trees that represent subsequent
Traditionally, accounting for the influence of actions, such as crew recovery. Other sources of
multiple shaping factors with multiple levels of information, such as the ASME Standard for
influence without imposing a high degree of expert Probabilistic Risk Assessment for Nuclear Power
consensus judgment on the HRA process has Plant Applications (2002), the PRA Procedures
proven difficult for HRA. SPAR-H attempts to Guide (NUREG/CR-1278 1983), or the NRC Fault
help make the assignment of human error Tree Handbook (Veseley et al. 1981), do address
probability a more repeatable function and less a these modeling issues in depth, including different
function of the analyst performing the HRA. We search strategies that can be used to identify
believe that the analyst’s expertise comes into play human failure events.
in discovery of the appropriate error and in
assigning the correct level of influence (i.e., We believe that the PSFs used in SPAR-H account
multiplier for the HEP). The HRA search process for most of the performance influences that will be
for determining unsafe acts given a particular observed in events and are the PSFs most
context still remains a challenging task for the applicable to support SPAR models. Potentially,
PRA/HRA analyst, but this is the information that there may be difficulty when reviewing an event in
is brought to SPAR-H for quantification. The need deciding the assignment of a particular influence
to provide sound qualitative assessments of factors to one of the PSF categories. For example, in the
is amplified as SPAR-H applications expand case of a preinitiator, poor work package
beyond basic plant PRA model development to development coupled with evidence of poor
include HRA for event analysis and the evaluation communications during pre-evolution briefing can
of specific plant performance issues. potentially increase the likelihood of error during
maintenance activities. In this instance the work
CAVEATS package development influence needs to be
mapped to the work practices PSF in SPAR-H. In
As does any simplified method, SPAR-H has almost all cases, different types of PSFs that
modeling and analysis limitations that should be appear in other HRA methods or that are
clear. We list several of these limitations here. developed through the analyst’s understanding of
SPAR-H does not address in detailed fashion how an operating event can be assigned to one of the
to incorporate SPAR basic events into system eight PSFs appearing in SPAR-H.
event trees. It does provide guidance, however, to
calculate or estimate the probabilities associated There may be rare cases in which it is difficult to
and the dependencies between those events once map the analyst assignments to the specific PSFs
the analyst has determined the appropriate system in SPAR-H. These cases are not a major concern,
model structure. SPAR-H offers a means for because SPAR-H encourages documentation of the
estimating the probability associated with recovery assumptions underlying PSF assignments. Three
attempts; the analysts can use the worksheet to considerations apply when mapping an analyst
explicitly model and quantify recovery attempts. A assignment to a PSF. First, the most important
person’s recovery from his or her individual error aspect of flagging a particular PSF is that the
is often the product of interface quality and analyst makes appropriate adjustments to the HEP,
systems feedback, procedures quality, training and that is, calls attention to the fact that a non-
experience, and supporting work processes. These nominal condition exists. If a specific PSF does
factors are reflected in the PSF assignment that not fully apply, the analyst should indicate this in
modifies the nominal rate. Functional systems the HRA analysis but nonetheless make
recovery as used in PRA is not recovery from an adjustments to the HEP. Second, assignment of
individual error per se but the restoration of extreme PSF values requires that the analyst have
function, and can involve many tasks. strong justification for the assignment, and that he
or she indicate the reasons for said assignment
within the body of the HRA analysis. Third, the

xix
effect of an individual PSF assignment may be versus postinitiator actions. The reason for this is
diminished, depending where the HEP appears in that a general model of human performance is
the fault tree or event tree structure and PRA assumed, and any differences noted in
model. In general, the HRA practitioner needs to performance can be accounted for through the
collect details on the scenario, including what proper application of PSFs. A priori, in his or her
information the crew needs, how they obtain this analysis, the HRA practitioner has indicated
information, and any factors that could interfere whether the HEP under evaluation is more suitably
with them reaching a proper diagnosis. addressed as a pre- versus postinitiator. People are
not different before and after an event except to
Although it would be preferable to have the extent that the event and its context provide
empirically derived PSF distributions, the feedback, raise or lower stress, call on operator
probability density functions (PDFs) employed in training and knowledge, interfere with information
SPAR-H make use of the same theoretical processing, reduce the availability of job
distributions that have been used in other HRA performance aids, influence complexity, or
methods. At the time that the SPAR-H work was otherwise affect PSFs that are defined in the
performed, we were not able to benchmark against method.
experimental or experiential data from the nuclear
industry. We conducted a review process, and the Although it would be advantageous to be able to
PSFs used in SPAR-H are supported by findings address uncertainty in more detail, at present it is
contained in the behavioral sciences literature. The difficult to determine the contribution of aleatory
PSFs are present in most other HRA methods, and versus epistemic sources of uncertainty on the
can be mapped to findings regarding the HEP as a function of PSF influences and
characterization of errors present in operating interactions. This is beyond the scope of a
events (see Gertman et al. 2002). The NRC is in simplified HRA method. However, we note
the process of examining human performance data numerous potential research avenues that could
on PSFs systematically collected by the Halden further our understanding in this area.
Reactor Project and reviewing similar information
taken from LERs under the NRC HERA HRA REFERENCES
database project under Job Code Number Y6123.
ASME RA-S-2002, Standard for Probabilistic
Last, SPAR-H allows for flexibility. Analysts can Risk Assessment for Nuclear Power Plant
decompose to different levels, as well as make Applications, American Society for
their own determinations regarding whether Mechanical Engineers, 2002.
important transitions such as those between Atwood, C. L., “Constrained Non-informative
procedures need to be individually identified in Priors in Risk Assessment,” Reliability
PRA system models. In the case of SPAR model Engineering and System Safety, 53, 1, pp. 37–
development; these transitions in procedure usage 46, 1996.
are not modeled as diagnostic basic events.
Typically, transitions are captured within a Bieder, C., P. LeBot, and F. Cara, “What Does a
system-level basic event. For example, the MERMOS Analysis Consist In?” PSA ’99,
operation of safety systems such as injection or Washington, DC, American Nuclear Society,
heat removal is represented as a single HEP for the pp. 839–845, 1999.
operation of that system, even though multiple Blackman, H. S., and J. C. Byers, ASP/SPAR
transitions may be required. If the transition Methodology, internal EG&G report
process is problematic and thought to influence developed for the U.S. Nuclear Regulatory
performance, the analyst employing SPAR-H has Commission, 1994.
the flexibility to increase the nominal rate by
assigning an appropriate non-nominal PSF level to EPRI TR-100259, Approach to the Analysis of
either complexity or to the procedures PSF. Operator Actions in Probabilistic Risk
Assessment, Palo Alto, California, 1992.
The current version of SPAR-H does not
differentiate in terms of PSF levels between pre-

xx
Gertman, D. I., et al., Review of Findings for Wakefield, D., G. W. Parry, and A. J. Spurgin,
Human Performance Contribution to Risk in Revised Systematic Human Reliability
Operating Events, NUREG/CR-6753, Analysis Procedure (SHARP1), EPRI TR-
Washington, DC, U.S. Nuclear Regulatory 10711, Electric Power Research Institute,
Commission, 2002. Palo Alto, California, 1999.
Hannaman, G. W., and A. J. Spurgin, Systematic Weil, R., and G. E. Apostolakis, “A Methodology
Approach to Human Reliability Analysis for the Prioritization of Operating Experience
Procedure (SHARP), EPRI NP-3583. Palo in Nuclear Power Plants,” Reliability
Alto, California, Electric Power Research Engineering and Systems Safety, 74, pp. 23 –
Institute, 1984. 42, 2002.
Hollnagel, E. Cognitive Reliability and Error Williams, J., “Toward an Improved Evaluation
Analysis Method (CREAM), Oxford, Elsevier, Analysis Tool for Users of HEART,”
1998. International Conference on Hazard
Identification and Risk Analysis, Human
NUREG-1624, Rev. 1, Technical Basis and
Factors and Human Reliability in Process
Implementation Guidelines for a Technique
Safety, January 15–17, 1992, Orlando,
for Human Event Analysis (ATHEANA),
Florida.
Division of Risk Analysis and Applications,
Office of Nuclear Regulatory Research, 2000. Veseley, W. E., et al., Fault Tree Handbook,
NUREG-0492, Washington, D.C., U.S.
NUREG/CR-1278, Handbook of Human
Nuclear Regulatory Commission, 1981.
Reliability Analysis with Emphasis on
Nuclear Power Plant Applications (THERP)
Final Report, Sandia National Laboratories,
1983.

xxi
xxii
 ACKNOWLEDGEMENTS

The authors wish to thank a number of individuals whose help greatly benefited this
report. From the national laboratory and industrial sectors these individuals include John
O’Hara, John Lehner, Marty Sattison, John Forester, Hugh Whitehurst, Alan
Kolaczkowski, Dennis Bley, and Steven Mays. EPRI staff members who supplied written
comments include John P. Gaertner, Gary Vine, and Nick Grantom. Industry
representatives present at public meetings who also supplied valuable comments include
Jeff Julius, Frank Rahn, and J. Grobbelaar.

From the INL, Ron Boring contributed to the latter versions of the document including
grappling with format and providing technical content, including contributions in the area
of cognitive science. Jennifer Nadeau and Lon Haney worked on earlier versions of the
report. Dave Pack provided technical editing support even with continual changes
including re-organization of the document and content.

NRC staff members were also a source of inspiration and provided a wealth of comments
that helped to enhance the final version. These members include Suzanne Black, James
Bongarra, Mike Cheok, Susan Cooper, Dave DeSaulniers, Mike Franovich, Claire
Goodman, Hossein Hamzehee, Chris Hunter, Paul Lewis, Erasmia Lois, David Loveless,
Gareth Parry, J. J. Persensky, Marie Pohida, Nathan Siu, Dave Trimble, and Peter
Wilson, as well as members of the ACRS PRA subcommittee, including George
Apostolakis, Dana Powers, Steve Rosen, and J. Sieber.

xxiii
xxiv
 ACRONYMS
AFWD auxiliary feedwater LB lower bound
AIT Augmented Inspection Team LCO limiting condition of operation
ASEP Accident Sequence Evaluation LDST let-down storage tank
Program LER licensee event report
ASME American Society of Mechanical LOI loss of inventory
Engineers LP/SD low power and shut down
ASP accident sequence precursor LTM long-term memory
ATHEANA A Technique for Human Event MERMOS Methode d' Evaluation de' la
Analysis Reaslisation des Missions
BWR boiling water reactor Operateur pour la Surete'
CAHR Connectionism Approach to MMPI Minnesota Multiphasic Personality
Human Reliability Inventory
CAP corrective action plan MOV motor-operated valve
CCDP conditional core damage MSIV main steam isolation valve
probability NASA National Aeronautics and Space
CCP centrifugal charging pump Administration
CN constrained non-informative NASA JSC National Aeronautics and Space
CNI constrained non-informative prior Administration Johnson Space
CREAM Cognitive Reliability Evaluation Center
and Analysis Method NPP nuclear power plant
CRO control room operator NRC Nuclear Regulatory Commission
CRS control room supervisor NSO nuclear service operator
CS containment sump PM plant management
CS core spray PRA probabilistic risk assessment
DG diesel generator PSF performance shaping factors
EF error factor Pwd probability (human error) with
EFC error forcing context dependency
EOC Emergency Operations Center Pw/od probability (human error) without
EOC error of commission dependency
EOP emergency operating procedure PWR pressurized water reactor
EPRI Electric Power Research Institute RCP reactor coolant pump
ESF engineered safety features RCS reactor coolant system
FLIM Failure Likelihood Index Method RHR (S) residual heat removal system
FMEA failure mode and effects analysis RI resident inspector
FSAR Final Safety Analysis Report ROP reactor oversight process
HEART Human Error Analysis and RPV reactor pressure vessel
Reduction Technique RX reactor
HEP human error probability SAPHIRE Systems Analysis Program for
HF human factors Hands-On Integrated Reliability
HF PFMEA human factors process failure Evaluation
modes and effects analysis SAR Safety Analysis Report
HFE human failure events SBCV Safety block control valve
HLR-HE-E High Level Requirements for SCUBA self-contained breathing apparatus
Human Error (ASME def.) SD shutdown
HMI human machine interface SG steam generator
HPI high-pressure injection SGTR steam generator tube rupture
HRA Human Reliability Analysis SHARP Systematic Human Action
INL Idaho National Laboratory Reliability Procedure
IPE individual plant examination SLIM Success Likelihood Index Method

xxv
SPAR standardized plant analysis risk TLX Task Load Index
SRV safety relief valve TOC Technical Operations Center
SS shift supervisor TRC time-reliability curve
STD Standard TS Technical Specifications
SM secondary memory TSC Technical Support Center
STM short-term memory UA unsafe acts
TH thermal hydraulics UB upper bound
THERP Technique for Human Error Rate
Prediction

xxvi
 GLOSSARY
Adjustment factor—The product of the when the method specifically uses that term), task
performance shaping factor (PSF) multipliers. The type, and error categories.
adjustment factor is only calculated when three or
more negative PSFs are present. The product is Event—A high-level generic term encompassing a
then used in the adjustment formula in conjunction non-normal occurrence at a nuclear power plant
with the nominal human error potential (HEP) to (or other facility).
produce the overall HEP. This helps to reduce
double counting of effects and restricts the Human error—An out-of-tolerance action, or
calculated mean value from being greater than 1. deviation from the norm, where the limits of
In situations where there are 2 or fewer negative acceptable performance are defined by the system.
PSFs, the PSF values are directly multiplied with These situations can arise from problems in
the nominal human error probability, and the sequencing, timing, knowledge, interfaces,
adjustment factor is not used. procedures, and other sources.

ASP SPAR (1994)—Accident Sequence Precursor Human error probability (HEP)—A measure of
Standardized Plant Analysis Risk; includes the likelihood that plant personnel will fail to
original iteration of SPAR-H, with following initiate the correct, required, or specified action or
characteristics: Process and diagnostic task response in a given situation, or by commission
distinction, no uncertainty information beyond will perform the wrong action. The HEP is the
adoption of error factors typically used in other probability of the human failure event (ASME
methods, Swain quantification approach to RA-S-2002).
dependency.
Human failure event (HFE)—A basic event that
Basic event—The term used in this report to represents a failure or unavailability of a
describe a component failure, loss of function, component, system, or function that is caused by
unavailability, or failed human action in a SPAR human inaction or an inappropriate action (ASME
model event tree. An example of a basic event RA-S-2002).
might be “Operator fails to throttle high-pressure
injection (HPI) to reduce pressure.” Initiating event—In the SPAR model
terminology, one of the high-level scenarios under
Error mode—Error type is also referred to as study (e.g., steam generator tube rupture, loss of
error mode. Major categorization schemes feed water, loss of offsite power, etc).
associated with first-generation methods include
omission or commission that can occur within the Joint HEP—In SPAR-H, a basic human failure
skill-, rule-, and knowledge-based domains. event (HFE) that has both diagnosis and action
Second-generation methods use terminology such parts. In preinitiator situations, this could include a
as slips, lapses, and mistakes, where the latter have task such as “trouble shoot and correct.” A post-
a large cognitive component that is accounted for initiator basic event could include “operator
through the analysis of context. The SPAR-H recognizes the need to energize systems before
method uses action and diagnosis as the major implementing the correct configuration and then
type tasks, and various error types are takes the appropriate action.” The resulting basic
distinguished. event is then reviewed for dependency and
modified accordingly.
Error type—The term used in this report to refer
to categories of human tasks. Other terms that are Low power and shutdown (LP/SD)—A set of
often used for this purpose are error mode, which nuclear power plant (NPP) operating modes,
is used in this report for describing specific human determined by an individual plant’s Technical
reliability analysis (HRA) methods (and then only Specifications (TS). However, most plants have
adopted, or are in the process of adopting, the

xxvii
NRC-approved Technical Specifications factor (PSF) definitions, influence factors and
associated with the various plant vendors. In range of influence determined by review of
pressurized water reactors (PWRs), there are six literature and HRA methods.
operating modes. In LP/SD PRA, Modes 4, 5, and
SPAR-H method (2004 revision)— Standardized
6 (which are subcritical) are reviewed. Mode 4
Plant Analysis Risk-Human Reliability Analysis
refers to hot shutdown; Mode 5 refers to cold
method; third (current) iteration of SPAR-H, with
shutdown; and Mode 6 is associated with
following characteristics: Action versus diagnosis
refueling. In a boiling water reactor (BWR), there
task distinction preserved, time influencing factor
are five operating modes. Modes 3, 4, and 5 refer
re-defined for low power and shutdown events,
to hot shutdown, cold shutdown, and refueling,
dependency refined, uncertainty calculation
respectively.
methods determined, ASME Standard for PRA
requirements addressed, clarification on recovery
Negative PSFs—In SPAR-H, negative
presented, at power and LP/SD considerations
performance shaping factors (PSFs) are those PSF
made explicit.
values that increase the nominal value rate, i.e., the
PSF values are greater than 1, are referred to as Subtask—In this report, a human action at a level
negative PSFs and figure in conjunction with lower than a task (i.e., basic event) level. May
positive PSFs in the overall HEP calculation. also be called a subevent.
When the number of negative PSFs is three or
Task—In this report, often refers to the human
greater, then the HEP adjustment factor is applied.
action(s) described in a SPAR model basic event
Performance shaping factor (PSF)—A factor [e.g., failure to recover residual heat removal
that influences human performance and human (RHR)]. The level of these tasks often
error probabilities is considered in the HRA encompasses relatively large numbers of human
portion of the PRA. In SPAR-H, this includes: actions, which might, in other circles, be called
time available, stress/stressors, complexity, tasks in their own right.
experience/training, procedures, ergonomics/
human-machine interface, fitness for duty, and Unsafe Actions—Those actions taken or omitted
work processes. that lead the plant into a less safe state. Only a
subset of human errors result in unsafe actions.
SPAR-H method (1999 revision)—Standardized Also, only some portion of unsafe actions lead to
Plant Analysis Risk-Human Reliability Analysis human failure events defined in the PRA model.
method; second iteration of SPAR-H, with For example, timing and available barriers may
following characteristics: Action versus diagnosis limit the number of unsafe actions that become
task distinction, changes in performance shaping human failure events.

xxviii
 1. INTRODUCTION
1.1 Overview of accident precursors (NUREG/CR-4674 1992).
The PRA models developed under the NRC’s
The Standardized Plant Analysis Risk (SPAR) Accident Sequence Precursor (ASP) program
human reliability analysis (HRA) method is a included aspects of HRA; however, the HRA
simplified HRA approach intended to be used in involved was not developed fully. This specific
conjunction with the development of SPAR method was designated the ASP HRA
probabilistic risk assessment (PRA) models. The methodology. Although, this original approach
language included in this document often refers to was adequate for a first generation of SPAR
aspects of SPAR models such as initiating events models concerned with screening analysis, the
and basic events—terms common to PRA. The NRC staff analysts decided that further refinement
glossary of this report presents general definitions of the HRA method was warranted and that this
for these terms. The SPAR-H method can also be effort should coincide with efforts underway to
used to support event analysis. This aspect of the refine the SPAR models. As a result, the Idaho
method is reviewed in Section 4.2.4. National Laboratory (INL) undertook a review in
1994, during which time a number of areas for
The process of carrying out HRA assumes that
improvement were noted. For example, in 1994
human error can be identified, modeled
the ASP HRA methodology was compared on a
(represented), and then quantified. Guidance for
point-by-point basis to a variety of other HRA
satisfying these requirements, including the
methods and sources. A team of analysts at the
process for error identification of events for
INL evaluated the differences among the methods.
inclusion in PRA models, may be found in IEEE
This evaluation led to a revision of the 1994 ASP
STD 1082 (1997) or the ASME Standard for
HRA methodology to incorporate desirable aspects
Probabilistic Risk Assessment for Nuclear Power
of these other methods. In addition, the revision
Plant Applications (ASME-RA-S-2002). We
also focused on addressing user comments.
assume that the human error probabilities (HEPs)
generated from the SPAR-H method will be used By 1999, the field of HRA had changed enough to
in PRA logic modeling structures, such as event cause the NRC to undertake a second revision to
trees and fault trees 2, so that there is a context the ASP HRA methodology. A revised
regarding how these estimates are to be combined methodology, named the SPAR-H method, was
and their effects interpreted. Modifying failure developed, and ASP was omitted from the title. A
probabilities based on dependency without regard revised form for applying the SPAR-H method,
to how the HEPs are to be combined can result in the SPAR Human Error Worksheet, was
erroneous conclusions about their potential developed and underwent testing by NRC
contribution to risk. inspectors. After using the method for a period of
time, a number of areas for improvement were
1.2 Background identified. These included more refined concepts
and definitions and suggestions for enhancing ease
The HRA approach presented in this document has
of use.
its origin in some of the early U.S. Nuclear
Regulatory Commission (NRC) work in the area At that time, the NRC’s Office of Nuclear
Regulatory Research identified two further areas
2
A fault tree is used to depict how component level for refinement. The first refinement involved
failures propagate to cause an undesirable system better assistance to the analyst, with understanding
failure (event). The system level event is the top-event or estimating the uncertainty associated with HEP
of the fault tree. Fault tree analysis offers a graphical estimates produced with the method. As an artifact
tool for understanding all those combinations of of the method’s early reliance on error factors,
component failures resulting in a specific system analysts could routinely produce upper-bound
failure. It is also useful in understanding how a
probabilities greater than 1 when modeling
particular component failure can be the result of lower-
level failures.
strongly negative performance shaping factors

1
(PSFs). This problem was not unique to part of the event analysis process performed by
performing SPAR-H. Although HRA analysts NRC staff, the method can also be used to support
have worked around this problem for 20 years, the detailed screening analysis whose goal can be the
INL was tasked to attempt to develop an easy-to- exclusion of human interactions from more
use but more suitable approach to representing detailed and complex HRA analysis. The SPAR-H
uncertainty information for use in analysis with the method differs from less detailed HRA in that it
SPAR models employing Systems Analysis requires analysts to consider dependency and a
Programs for Hands-on Integrated Reliability defined set of PSFs when performing
Evaluation (SAPHIRE) software (NUREG/CR- quantification. For example, analysts using
6618 2000). techniques such as the Failure Likelihood Index
Method (FLIM) or the Success Likelihood Index
The second refinement involved the applicability Method (SLIM) are free to include any number of
of this approach to support NRC-sponsored model PSFs that they think might apply. The SPAR-H
development research in the area of low power and method also differs from some of the earlier time-
shutdown (LP/SD) risk analysis. Specifically, reliability curve (TRC) methods in that the SPAR-
inquiry was made regarding whether the method, H method does not overly rely on time as the
as configured, was easily applied to LP/SD primary determinant of crew performance, but
scenarios. When the SPAR-H method was first rather treats time as one of a number of important
developed, there were no SPAR models for LP/SD shaping factors influencing human performance.
and, at that time, the HRA analysts had not
considered LP/SD as constituting a separate class SPAR-H also does not explicitly distinguish
of events that could require either subtle or major among skill-, rule- and knowledge-based
adjustments to the method. behaviors. Extended TRC data collection
encompassing 1,100 simulator trials failed to
1.3 HRA Orientation verify the independence of these definitions (see
EPRI TR-100259, 1992).
The goal of HRA is to support PRA in identifying
and assessing risks associated with complex 1.3.1 Guidance in performing HRA
systems. PRA, in conjunction with HRA, affords
A number of guidance documents are available
analysts the ability to look at sequential as well as
that can be used to support the SPAR-H method.
parallel pathways that generate risk, including the
These include the IEEE Guide for Incorporating
human contribution to that risk. Insights are gained
Human Action Reliability Analysis for Nuclear
by applying event frequencies to hardware failure
Power Generating Stations (IEEE STD 1082,
models and reviewing expected frequencies for
1997), Systematic Human Action Reliability
various hazardous end-states by condition
Procedure (SHARP; Hannaman and Spurgin
assessments.
1984), and the ASME Standard for Probabilistic
From the authors’ perspective, HRA is performed Risk Assessment for Nuclear Power Plant
as a qualitative and quantitative analysis. It helps Applications (ASME RA-S-2002). The IEEE
the analyst to study human system interactions and recommended practice for conducting HRA (IEEE
to understand the impact of these interactions on STD 1574) is under development and when
system performance and reliability. The SPAR-H completed will also provide a framework for
method is used to assist analysts in identifying conducting HRA.
potential vulnerabilities. The SPAR-H method can
We assume that a number of principles suggested
also be used to characterize preinitiating actions,
in these various references will be adhered to,
initiating event-related actions, and postinitiating
including the following:
event interactions. The SPAR-H quantification is
used because it is an efficient and not overly time • Identify and define the scenario or issue of
consuming approach to representing human interest.
actions and decisions in the final SPAR analysis
• Review documentation when possible,
model. Although the SPAR-H method is used
including event and near-miss databases,
primarily in SPAR model development and as a

2
 procedures, and the Safety Analysis Report Section 3 summarizes and discusses the approach
(SAR). and compares this HRA method against some of
the criteria for HRA, as defined by the new ASME
• Perform limited task analysis—walk down
PRA Standard, and the PRA Procedures Guide for
systems, conduct interviews, review
NASA Managers and Practitioners (Stamatelatos
appropriate training materials, and review
and Dezfuli 2002). Last, this report contrasts the
videotape and crew simulator performance.
SPAR-H method against criteria developed by the
• Screen and document—build a qualitative authors for review of HRA methods (in Gertman
model integrated with systems analysis. and Blackman 1994).
• Quantify. Section 4 presents considerations when using the
• Perform impact assessment. SPAR-H method, reviews application of the
SPAR-H method to event analysis, and addresses
• Identify and prioritize modifications to reduce use of the SPAR-H worksheets.
risk.
Section 5 presents consideration of PSFs for at-
• Document. power and LP/SD scenarios, examines results of a
sample application of at-power, and LP/SD
1.4 Organization approaches to a loss of inventory (LOI) scenario,
and reviews base error rates for diagnosis and
This report is archival, that is, it contains historical action tasks.
information regarding SPAR-H method
development, as well as provides an overview, Appendices A and B present SPAR-H worksheets
review of technical basis, and sample applications used to support the analyst application of the
of the method. Section 1 presents the background method to at-power and low-power/shutdown
and general HRA approach. (LP/SD) scenarios. Appendices C and D present
at-power and LP/SD examples, respectively.
Section 2 details the information processing-based Appendix E presents SPAR-H results for dry cask
model from which the SPAR-H method was risk assessment. Appendix F presents operational
developed. Summary performance influencing examples for the SPAR-H assignment of PSF
factors are introduced; task and error types are levels. Appendix G shows the relative relationship
defined; and the relation of SPAR-H PSFs to other among SPAR-H PSFs. Appendix H presents the
HRA methods is discussed. The approach to SPAR-H development history. Last, Appendix I
dependency and uncertainty factors, including presents a compendium of SPAR-H review
quantification, is also reviewed. comments from external review and public
meeting forums.

3
4
 2. SPAR-H METHOD
2.1 Model of Human stimulus-response and the information processing
approaches. This is because the HRA analyst
Performance needs to be able to consider aspects of diagnosis
Models of human behavior are discussed in a and planning as well as the likelihood of the
variety of behavioral science sources that deal with operators’ ability to successfully carry out actions
cognition [see, for example, Anderson (1995); often identified through procedures. This
Medin and Ross (1996)]. The cognitive and distinction between diagnosis (i.e., information
behavioral response model developed for the processing) and action (i.e., response) is the basis
SPAR-H method was developed out of early for separate diagnosis and action worksheets,
cognitive science approaches and is generally contained in Appendices A and B, with separate
termed an information processing approach to probability calculations.
human behavior. The factors constituting the basic
elements of this model also come from the SPAR-H also acknowledges the role of
literature surrounding the development and testing environmental factors upon diagnosis and action.
of general information processing models of For example, during evaluation of performance
human performance. Most information processing shaping factors, analysts note whether interactions
models of human behavior include representation might be difficult to analyze due to misleading
of perception and perceptual elements, memory, indications, complexity, time-dependent aspects,
sensory storage, working memory, search strategy, and the effects of combinations of unavailable or
long term memory, and decision making (see faulted equipment. Components of the SPAR-H
Sanders and McCormick, 1993). behavioral model approach, presented in
Figure 2-1, are discussed below.
Other psychological models or paradigms such as
stimulus-response models have been developed to Information flow from the environment can be
aid in understanding human behavior. In the across different sensory modalities: visual,
stimulus-response approach much of cognition is auditory, and kinesthetic. Environment factors can
not considered; rather, reflexive behavior is act to filter this information. Perhaps the easiest
developed over time as a function of learned example of this is how noise in the environment
associations between human actions and rewards can operate to mask the strength of an annunciator.
or punishments. Equipment response characteristics can also alter
the strength or nature of available sensory
The SPAR-H model combines elements of the information. This is present in phenomena such as

Human Behavior Model

Individual Factors

Figure 2-1. Human Performance Model.

5
speech clipping when using certain types of can vary depending upon whether meaningful
communication equipment. Still other filters information can be chunked, i.e., grouped, or not.
internal to the operator exist as well. These We tend to view short term memory as a process
include the influence of language, experience, and through which information is available for use by
expectancies. cognitive processes. Hence, both long term and
short term memory play a role in a human
Perception can be simple and direct. We refer to information processing model.
this as detection. An example is when an operator
detects that a low level alarm has actuated or Short term memory has been reinterpreted by
detects that there is a change in a trend plot or others. For example, Shallice (1982) describes a
other charting device. This perception acts as a supervisory attention model with limited capacity
bridge between physical sensation and cognition. that is directed toward cognitive tasks such as
Aspects of detection include identification and decision making and planning, and where
recognition, which are also influenced by these attentional resources are directed on the basis of
filters. Bodies of research have focused upon the degree to which the task involves novel or
detection under different conditions. The more technically difficult situations such as those where
well known of these are referred to as studies in strong habitual responses or behaviors may be
signal detection and are reflected in signal inappropriate. Baddeley (1990) describes a
detection theory. Other research in the behavioral working memory model that includes a central
sciences has examined the role of experience, executive similar to the Shallice model. This
learning, training, and beliefs upon perception and central executive invokes, directs, and integrates
perceptual processes. SPAR-H incorporates these processing routines with supervisory attention
mechanisms at a very high level via the phenomenon described in Shallice (1982).
assignment of performance shaping factors. Baddeley’s expanded model includes a visual
sketchpad component that enables temporary
Aspects of high order information processing storage and manipulation of spatial and visual
present in the SPAR-H approach consist of short information. There is also a phonological loop,
term, external, and long term memory. which is responsible for manipulation and
McCormick and Sanders (1993) view the human temporary storage of auditory/verbal information.
memory system as being based upon three The SPAR-H underlying model is a simplified
processes. These are sensory storage, short term memory framework akin to the Anderson (1995)
memory, and long term memory. These processes approach. It may, however, also be interpreted
work on two types of memory systems, auditory according to the more detailed model of human
and visual. Evidence on the existence of these two memory documented in Baddeley (1990) and
distinct sensory memory systems is reviewed in others.
Anderson (1995). He notes that there is an iconic
memory for visual information processing and an The SPAR-H model also includes external
echoic memory for auditory processing. SPAR-H memory, which consists of information that a
acknowledges these components of memory but person may use to aid their short and long term
does not model them explicitly as part of the HRA memory. Examples of external memory are the
process. different types of operating procedures, in which
the steps of a task are enumerated for reference by
For a through review of this area of human the operator. The operator does not need to retain
performance the reader should refer to Anderson this information in short or long term memory.
(1995). For example, existence of a brief visual Rather, the information is available to reference
sensory store can be traced back to Sperling’s whenever the operator needs it. In SPAR-H,
research in the early 1960s. In the underlying external memory is modeled as the performance
SPAR-H model, short term memory can be shaping factor for procedures.
construed as the ability of the operator to keep a
limited amount of information in an active mental Demand characteristics of the task impact the
state. Long term memory items must be activated internal resource requirements of the operator. For
and retrieved. The capacity of short term memory example, tasks that require the operator to perform

6
mental calculations or maintain multiple Working memory and short term memory model
hypotheses while keeping track of other tasks or to aspects are based upon factors including capacity,
perform monitoring functions reduce the available rehearsal, and attention. Memory capacity is
mental resources, thereby leading to error. High physically fixed, but training can make operators
demand has been shown to interfere with recall. more effective at chunking information, thereby
Physically demanding tasks can also deplete increasing the storage efficiency of memory.
internal resources, producing fatigue that can Rehearsal refers to the use of memorization,
result in higher than expected human error for training, and operations experience, which can aid
physically demanding and cognitively demanding in the speed and ease of retrieving memories by
tasks. keeping information active in memory. Attention
is directed and influenced by stress, task and
Environmental and situational factors are environment complexity, experience, and training.
contributors to the success or failure of human Attention is further directed by procedural cues.
performance via their impact upon perception,
processing, and response. High levels of For example, procedures, determined to be an
complexity, e.g., ambiguous problems involving influencing factor in operating events, also have a
multiple faulted systems, more than one solution, basis in information processing as an external
and producing unsuspected interactions, can result memory aid. Procedural errors or inadequacies in
in cognitive overload where perception, processing format and lack of appropriate cautions or
and response are compromised. High complexity warnings can increase the likelihood of human
interferes with short term and long term memory error. Lack of procedures or manuals can directly
components. For example, system relationships increase this likelihood. In the past, procedure
may be relatively complex, and the configuration problems, such as assumptions by the procedure
and flow of events are not well recognized. The writers that discharge high-pressure injection
operator may not be able to recognize the true (HPI) pressures would be available to control
nature of the problem and, thus, is challenged to room personnel when they were not contributed to
determine a solution from memory. In this decrements in human-system performance (e.g.,
situation, it would be more difficult to determine the Oyster Creek event in U.S. NRC 1992).
what was occurring and to take the correct course Procedures also can interact with other work
of action. The SPAR-H analyst represents practice factors such as quality assurance or
complexity and directly by assignment of the training. For example on May 3, 1997, during the
appropriate PSF level. Higher levels of complexity shutdown process at Oconee Nuclear Station, Unit
are assumed to be associated with greater human Three (U.S. NRC 1997), a single reference leg for
error. both channels of the let-down storage tank (LDST)
instrumentation was being used without caution in
Table 2-1 presents operational factors in SPAR-H procedures for common cause failure. This
that are mapped to the information and behavioral contributed to a reactor trip. Also, procedural
model discussed above. Review of the behavioral inadequacies in place at the time allowed for the
sciences literature reveals eight summary LDST level to be maintained in a range lower than
operational factors, or PSFs, associated with the alarm setpoint.
nuclear power-plant operation. These operational
factors can be directly associated with the model Evidence for the importance of workload in human
of human performance. Within the table, various performance has been established through the use
aspects of performance and their relation to the of primary task measures and secondary task
PSFs are indicated. For example, perception is measures. At first, psychological studies used the
limited based upon human sensory limits, is relatively crude time reliability approach used in
susceptible to disruption or interference, and First Generation HRA. Workload was defined in
occurs as a function of modality (auditory, visual, terms of the ratio of the time available to the time
or kinesthetic). Perception by operators is often a required to perform a task. It was soon realized
function of the quality of the human machine that this approach fails to distinguish between
interface (HMI). tasks that can be successfully time-shared and

7
Table 2-1. Operational Factors in SPAR-H

(The numbers after each entry refer to the PSF list at the bottom of the table.)
Working Memory/ Processing and
Inflow and Perception Short-term Memory Long-term Memory Response
6,3
Presence (is the signal there?) and Limited capacity5 Training4 (models, problem solving, Training (actions)
4

opportunity (is anyone present to receive the *Serial processing behaviors) *Existing models of behavior
signal?) *Good only for a short time 2,3,5,4 • learning *Practice and skill
(20 seconds)
Human sensory limits2,5,7 Experience4 (models, problem solving, Experience 4 (actions)
behaviors) • practice and skill
Modality 6,5 (verbal, graphic/symbol, text) • learning • existing models of behavior
• echoic Right amount of attention 2,3,4,5,7
• iconic required Culture 8 (societal, organizational, Proper controls available6
• kinesthetic interpersonal, (crew))
Rehearsal 2,3,5,7 • learning Human action limits6,7 (physical strength
Physical and mental health 7 and sensory acuity)
Interference 6,5,4,7 (signal, noise) Intelligence/cognitive skills3,4,1,5,7 (decision
making, problem solving) Ergonomics of controls6,3
complexity
7
Interference factors 6,2,3, (distraction)
Environmental degradation2,3,6
1,3
Available time
Time to react versus time available 1
Physical and mental health 7

Performance Shaping Factors

Available time1 Procedures (including job aids)5

Stress and stressors 2 Ergonomics and human-machine interface6
Complexity 3 Fitness for Duty7
Experience and training 4 Work processes8

Note: Available time, from the operator’s perspective, is influenced by information complexity, which can take more processing and reduce the time
available to act.

8
those that cannot. Nor does it account for multiple 2.1.1 The Role of Work Processes
resources (both internal and external to the
Work processes are present in the model described
operator) or explain how situations could exist
above in terms of the management and
where the performance was identical and yet the
administrative environments parameter of the
one task was more demanding. Research in
model and are present in the “work processes”
secondary task performance, as a means to assess
PSF included in the SPAR-H worksheets
spare capacity on the part of subjects, has been
(Appendices A and B). The influence of work
used to assess the workload associated with
processes in operating events has been recently
performance on the primary task. Although large
highlighted. For example, a review of 37 operating
numbers of studies have used subjective measures
events at U.S. commercial nuclear power plants
to assess stress, physiological measures have also
(NPPs) from 1991 through 1999, conducted for the
been used to assess the stress associated with
NRC’s Office of Nuclear Regulatory Research,
mental workload. There is ample evidence of
revealed a number of instances where work
workload influence on performance.
processes affected crew demands during operating
The SPAR-H approach acknowledges the role of events (Gertman et al. 2002). The errors and
workload in influencing performance in a failures that occurred in these events included
quantitative way through PSF assignment to deficiencies related to design and design change
complexity and stress. Generally speaking, the work practices (81%), inadequate maintenance
effects of physical workload such as having to practices and maintenance work controls (76%),
perform multiple tasks, complete tasks more and corrective action program inadequacies (38%).
quickly, or move objects of increasing weight are
Work process factors were also implicated at the
captured under the PSF for stress and stressors; the
Dresden 3 event, May 15, 1996 (US NRC 1996),
effects of cognitive workload such as having to
where failure of a feedwater regulating valve led
perform additional calculations, refer to multiple
to a reactor trip. Work process involvement in this
sources of information to verify readings, or co-
event includes: inadequate inspection frequency of
ordinate actions based upon periods of waiting are
the feedwater regulating valve, and failure to
captured under complexity.
challenge generic work package inadequacies.
In general, the SPAR-H model is a structural Failure of the PCS relay was attributed to licensee
model for guiding analysis as opposed to a failure to trend repair information, inordinate delay
mathematical model of human information to placing FRV back into service, and failure to
processing. The SPAR-H basic information- follow industry practices resulting in control
processing model can be used to aid switches being in the wrong position prior to
conceptualizing the key aspects of an information- resetting Group 1 isolation.
processing model of human performance, which
Recently, the root cause analysis report for
reflects psychological principles. The purpose of
documenting the degradation of the reactor
beginning with this model is to account for and
pressure vessel head corrosion incident at Davis
integrate the factors key to human performance
Besse (2002) identified a number of work process
when performing SPAR-H analysis.
or organizationally bound factors that may have
In addition, these operational factors can be linked contributed to the event. Implicated were a flawed
to the portion of the human information-processing boric acid corrosion control program, and
model with which they are associated. The relation subsequent failures such as lack of written
of summary factors to information processing evaluations, inadequate implementation of utility
model parameters is presented in Table 2-1. The corrective action program, and lack of safety
model is also useful in terms of presenting the analysis for identified conditions. The root cause
basis for how operational factors impact team also concluded that there was failure to take
performance. Definitions of these eight PSFs actions for identified adverse conditions, failure to
follow in Section 2.4.4. trend, and failure to provide adequate training to
personnel. These factors point toward inadequate
work processes and inadequate implementation of

9
work processes. See also the Augmented components whose status can be changed to
Inspection Team Report (U.S. NRC 2002). reduce or eliminate the problem. It includes
interpretation and (when necessary) decision
For a more in-depth review and approach to work making. Diagnosis tasks typically rely on
process evaluation, see Weil and Apostolakis knowledge and experience to understand existing
(2002) who present a series of work process conditions, plan and prioritize activities, and
parameters and modeling approach to work determine appropriate courses of action.
process issues.
When answering the question “Does this task
2.2 Task Types contain a significant amount of diagnosis
activity?” one should consider whether the
The 1994 ASP HRA methodology divided tasks
operator or crew has to expend mental energy to
performed by personnel into two components, the
observe and interpret what information is present
processing component and the response
(or not present), determine what that means, think
component. Comments received from those trying
of possible causes and decide what to do about it.
to implement the method indicate this “processing
The greater the amount of observing, interpreting,
and response” delineation was understood by
thinking and deciding the operator or crew
human factors and HRA professionals working on
performs, the more significant the amount of
the method but proved difficult for trainers,
diagnosis activity that is taking place.
operators, and inspectors who were collaborating
on its application. 2.2.2 Guidance for Action
In 1999, these components were renamed in the Guidance for action has to do with carrying out
SPAR-H method as “diagnosis” and “action.” one or more activities (e.g., steps or tasks)
Comments received suggested that this separation indicated by diagnosis, operating rules, or written
of task types was more easily understood. This procedures. Examples of action tasks include
represents a top-level distinction between tasks operating equipment, performing line-ups, starting
that are often used in HRA (some applications also pumps, conducting calibration or testing, carrying
classify actions as preinitiator, initiator-related, or out actions in response to alarms, and performing
postinitiator). other activities during the course of following
plant procedures or work orders.
Within comments and task description fields of the
SPAR-H worksheets, the SPAR-H method allows 2.2.3 Guidance for Diagnosis and Action
analysts to use more complete descriptions for
tasks. However, quantification is based on the In performing HRA, it is sometimes practical and
assignment of tasks to one of two types, diagnosis reasonable to model on the basic event level a task
or action. In some ways, this simple delineation is or subtask containing aspects of both diagnosis
close to THERP in how it assigns tasks to support and action. In these situations, both diagnosis and
quantification. When using this approach, action portions of the HRA worksheet are to be
activities such as planning, intra-team applied. For example, consideration of the task
communication, or resource allocations during “operator terminates loss of inventory (LOI)
event progression are considered diagnosis. during LP/SD” as part of the event tree for LOI
should be modeled as a joint HEP. In order to
When using SPAR-H, the analysis team makes terminate loss of inventory, the operator or crew
decisions regarding the assignment of a particular must diagnose, that is, identify the leak path (the
post- or preinitiator activity to either diagnosis or difficulty of which may vary, depending on
action. whether the operators in the control room
contributed to plant conditions and can simply
2.2.1 Guidance for Diagnosis
retrace their steps to observe the error or whether
Guidance for diagnosis has to do with attributing actions independent of the control room
the most likely causes of the abnormal event to the contributed to the leak path and are more
level required to identify those systems or challenging for the operators to determine). The

10
subsequent action(s) to be taken, either inside or requiring diagnosis are evaluated on the action
outside the control room, are then evaluated on the task worksheets. When comparing PSFs among
worksheets. Both the diagnosis and action HRA methods, INL matched specific PSFs based
elements of the task are then quantified as part of upon appropriate diagnosis and action distinction
the SPAR-H HEP determination process. components is reflected in the two comparison
matrices developed (one for diagnosis, one for
In a number of situations, arguably there will be a action). These matrices are presented in this report.
dependency between the diagnosis and action
elements representing a single basic event. In If any PSFs, either from the 1994 ASP HRA
SPAR-H, the intradependency within a single method or from the other HRA methods, did not
basic event consisting of action and diagnosis is match, they were noted. The team reviewed these
acknowledged when combining (adding) these two nonmatching PSFs and other observations from
elements to yield the composite or joint HEP, i.e., the initial, nonquantitative comparison process,
basic event value. Interdependencies between and used this information to assist in developing
previous and subsequent basic events (HEPs) are the SPAR-H method and associated PSF
calculated according to the dependency portion of definitions. The objective for the completed
the worksheet in Appendixes A and B. version of the SPAR-H method was to cover the
important shaping factors noted in these methods.
In order not to be overly conservative, SPAR-H These PSFs are present in the human performance
suggests that analysts use the HEP adjustment model presented in Section 2 (Figure 2-1).
formula when calculating the HEP for those
situations where at least three HEPs are 2.3 Error Types
determined to be nonnominal and negative (i.e.,
the value of the PSF is greater than 1). Review of In a manner similar to the PSF matching
the dependency assignment process is presented in performed as part of the SPAR-H method
subsequent sections of this document. development process, the base error types from the
other HRA methods were compared with the 1994
In general, there are better-established HRA data ASP HRA method error types. This comparison
and data sources for actions than there are for was considerably easier than the PSF matching. It
diagnosis and planning activities. If cognitive was easier because it was straightforward to judge
activities are modeled and quantified with the whether or not other error types corresponded to
SPAR-H method and determined to pose a either or both of the processing and response error
significant contribution to risk, then analysts’ types of the 1994 ASP HRA method, as well as the
evaluations should employ a more detailed HRA new terminology (diagnosis or action). Early
method. These methods include: a technique for versions of the ASP HRA method attempted to
human event analysis (ATHEANA; NUREG-1624 differentiate between errors of omission and errors
2000); Methode d' Evaluation del' la Reaslisation of commission. Experience demonstrated that this
des Missions Operateur pour la Surete' distinction was not useful in making more accurate
(MERMOS; Bieder et al. 1999); or the predictions of error.
connectionism approach to human reliability,
(CAHR) Strater (2000). Where available, the EPRI Therefore, for the base failure rates(s) for
ORE/Cause Based Decision Tree (CBDT) Method diagnosis and action, the SPAR-H method uses a
(EPRI TR-100259, 1992) may also be consulted. composite rate for omissions and commissions.
In subsequent sections of this report, these Since the first ASP HRA method screening
methods are discussed briefly and compared with version, the discussion of omission and
the current SPAR-H method. commission within the HRA community for
describing error has slowly moved toward such
If the SPAR-H method is being used to evaluate a terms as slips, lapses, and mistakes. This, in part,
basic event consisting of multiple actions and is due to intuitively appealing evidence that there
decisions, such as is often the case with SPAR is an important difference between slips and
models, both diagnosis and action worksheets mistakes, the two frequently discussed errors of
apply. Tasks that are proceduralized actions not commission.

11
The first type of commission is properly called a base failure rates is documented in THERP. The
slip (i.e., right intention but wrong execution); the current generation of HRA methods, often referred
second is called a mistake (i.e., having a wrong to as second-generation HRA, also uses PSF
impression of what to do coupled with an information in one form or another when
improper action or decision). Review of the calculating HEPs. When assigning the PSF level,
context will help the analyst to determine whether the analyst evaluates the PSF from the perspective
slips or mistakes are more likely and whether these of the operator. Thus, the analyst would evaluate
errors are likely to have common cause the complexity of the diagnosis or action required
implications. Most second-generation HRA for a scenario or range of scenarios from the
approaches now emphasize that context, that is, perspective of the operator as opposed to the
combinations of PSFs, plant conditions, and analyst’s view of the complexity as a whole.
situational factors function together as a major
determinant of mistakes. The PSF emphasis in the In 1999, changes to the ASP/SPAR HRA method
SPAR-H method is intended to reflect incremental were implemented. The changes made at this stage
progress and direction in contemporary HRA. were in error type, PSFs, and in their definitions.
For example, the definitions associated with the
Thus, it is equally important, from a screening performance shaping factors became more
perspective, to be able to address PSFs that are expansive in nature to cover aspects of PSFs being
assumed to contribute to context, as it is to recognized in other methods. Also, some methods
distinguish among a slip, lapse, or a mistake. From distinguished between PSFs that were represented
a methodological perspective, it is important to by a single PSF in the SPAR-H approach.
emphasize that the HRA analysis team needs to
follow an approach that systematically identifies The changes were made based on field-testing and
those errors likely to result in unsafe acts, indicated that:
evaluates the influence of major PSFs, and
estimates their probability of occurrence. • “The raters don’t understand the processing/
response dichotomy”
The composite approach used in SPAR-H is also
believed to encompass other error taxonomies. For • “Most of the other HRA methods recognize
example, the nonuse of available information and separate diagnosis and action error types”
the incorrect use of available information as • “Other HRA methods have organizational
described by Hacker (1986) are assumed to be factors as a PSF.”
covered by the nominal rate. The presence of
technically inaccurate information is covered as In 1999, changes were also made to ensure the
well, and indicated by adjustment of the human SPAR-H method was as broad in coverage as
system interaction (i.e., ergonomics/HMI PSF). possible. Once the changes in error types and PSFs
During the error identification phase of the were made, new lists were created for error types
PRA/HRA process, nonroutine, significant errors and PSFs. Eight PSFs were identified: available
of commission that represent operator or crew time, stress and stressors, complexity, experience
mistakes should be considered for explicit and training, procedures, ergonomics and human-
modeling and quantification by the risk analyst. machine interface, fitness for duty, and work
processes. These same PSFs are present in the
2.4 PSFs 2004 version; they differ only in terms of their
description.
Many, if not most, HRA methods use PSF
information in the estimation of HEPs. In general, Next, comparison matrices were created (one for
PSF analysis enhances the degree of realism the new diagnosis error type, one for the new
present in HRA analysis. The extent and resolution action error type) that compared PSFs and their
of PSF analysis should only be specific enough to weight multipliers for SPAR-H method PSFs
identify potential influences and rate them on the
corresponding SPAR-H worksheets. Historically,
the first use of PSFs in HRA to modify nominal or

12
Table 2-2. HRA methods used in SPAR-H comparisons.
HRA Method Date Authors Focus - Purpose
1
CREAM 1998 E. Hollnagel Human performance classification based on error modes and consequences
(phenotypes) and causes (genotypes). Uses simple Contextual Control Model
(CoCoM) of cognition that includes continuous revision and review of goals
and intentions. Assesses cognitive function failures and common performance
conditions (CPCs) to support failure rate estimations.
HEART2 1988 J. Williams HRA based on nine generic tasks with individual nominal error rates. Analysts
identify error-producing conditions (EPCs). EPCs operate as multipliers to
increase base failure rates; their basis is in the behavioral sciences literature.
THERP3 1983 A.D. Swain and Developed to provide representational modeling of human actions (HRA Event
NUREG/CR-1278 H.E. Guttmann Trees) and estimation of HEPs. Emphasis is on nuclear power plant
applications to support PRA Provides HEP tables based on data gathered from
[Developed in the
various domains.
1970s and refined in
early 1980s].
ASEP4 1987 A.D. Swain Developed to provide an efficient method for estimation of screening HEPs for
pre- and post-accident human actions. Based on THERP.
NUREG/CR-4772
5
SHARP1 1990 Wakefield, et al Developed to provide a consistent approach to HRA assessments. Contains
performance shaping factor information. Addresses pre- and post initiator
conditions. Revision to early work in this area under the same name.
1
Cognitive Reliability and Error Analysis (CREAM) Method.
2
Human Error Analysis and Reduction Technique (HEART).
3
Technique for Human Error Rate Prediction (THERP).
4
Accident Sequence Evaluation Program (ASEP) Human Reliability Analysis Procedure.
5
Systematic Human Action Reliability Procedure (SHARP1)

13
Table 2-3. Action PSF Comparison Matrix, at power (PSFs = 8).
SPAR-H SPAR-H HEART CREAM
PSFs SPAR-H PSF Levels Multipliers Multipliers Multipliers ASEP Multipliers THERP Multipliers
Available Inadequate Time P(failure) = 1.0 P(failure) = 1.0 - Table 7.2 P(failure) = 1.0 - Table 20.1
Time
Time available = time 10 11 - EPC 5 - CPC 20 10 - Table 7.2 10 - Table 20.1
required
Nominal time 1 1 1 - CPC 19 1 - Table 7.2 1 - Table 20.1
Time available > 5 x .1
time required
Time available > 50 x 0.01 0.5 – CPC 18 0.01 -Table 7.2 0.01 - Table 20.1
time required
Stress/ Extreme 5 5 -Table 7.3 5, 25 - Table 20-16
Stressors
High 2 1.3 - EPC 29 1.2 – CPC 22 2, 5 - Table 20-16
1.15 – EPC 33
Nominal 1 1 – CPC 21
Complexity Highly complex 5 5.5 – EPC 10 2 – CPC 17 2.5 or 5 (depending on stress)
Moderately complex 2 1 – CPC 16
Nominal 1 1 – CPC 15
Experience/ Low 3 17 – EPC 1 2 – CPC 25 10 -Table 8.3 2 - Table 20-16
Training
3 – EPC 15
8 – EPC 6
6 – EPC 9
4 – EPC 12
2.5 – EPC 18
2 – EPC 20
1.6 – EPC 24
Nominal 1 1 1 – CPC 24 1 1
High 0.5 0.8 – CPC 23 0.1 - Table 8.3
Procedures Not available 50 P(failure) = 1.0 - Table 7.1, 50 - Table 20.7
Table 8.1
14
Table 2-3. (continued).
SPAR-H SPAR-H HEART CREAM
PSFs SPAR-H PSF Levels Multipliers Multipliers Multipliers ASEP Multipliers THERP Multipliers
Incomplete 20 5 – EPC 11 2 – CPC 14 10 - Table 20-7
3 – EPC 16,17
1.4 – EPC 28
1.2 – EPC 32
repeat
Available, but poor 5 5 – EPC 11 2 – CPC 14 10 - Table 20-7
3 – EPC 16,17
1.4 – EPC 28
1.2 – EPC 32
repeat
Nominal 1 1 – CPC 13

Ergonomics/ Missing/Misleading 50 P(failure) = 1.0 - Table 7-1, 100, 1000 - Table 20-12
HMI 8-1
Poor 10 10 – EPC 3 5 – CPC 11 6 - Tables 20-9, 11, 12
9 – EPC 4 2 – CPC 7 10 - Tables 20.10, 13, 14
8 – EPC 5, 7
4 – EPC 13, 14
2.5 – EPC 19
1.6 – EPC 23
1.4 – EPC 26
1.2 – EPC 32
repeat
Nominal 1 1 – CPC 9, 10, 6
Good 0.5 0.8 – CPC 5
0.5 – CPC 8

15
Table 2-3. (continued).
SPAR-H SPAR-H HEART CREAM
PSFs SPAR-H PSF Levels Multipliers Multipliers Multipliers ASEP Multipliers THERP Multipliers
Fitness for Unfit P(failure) = 1.0
Duty
Degraded Fitness 5 1.8 – EPC 22
1.2 – EPC 30
1.1 – EPC 35
Nominal 1
Work Poor 2 2 – EPC 21 5 – CPC 29
Processes 1.6 – EPC 25 2 - CPC 4
1.4 – EPC 27 1.2 – CPC 3
1.2 – EPC 31 1 - CPC 28
1.06 – EPC 36
1.03 per add’l
man – EPC 37
Nominal 1 1 – CPC 2,27
Good 0.8 0.8 – CPC 1
0.5 – CPC 26

16
versus PSFs and multipliers for other complexity situations’ potential influence on
contemporary HRA methods. These results are performance. The SPAR-H method assigns a
presented in Table 2-2. multiplier of 5 for complex situations, whereas
HEART assigns 5.5, and CREAM a 2. THERP
As part of this PSF comparison process, four does not treat complexity as a separate PSF.
contemporary PSF-intensive methods were However, recent methods such as CAHR
selected by HRA analysts for comparison. These (Strater 2000) point out the importance of this PSF
other methods were HEART, CREAM, accident as a determinant of behavior.
sequence evaluation program (ASEP) and THERP.
Only one, ASEP, approximates a screening level Experience and training effects are well
approach. The others may be used to support a documented in the behavioral sciences and
detailed HRA analysis. The comparison between training literature. The range of effect for this
the SPAR-H method and individual HRA methods particular PSF is relatively large, ranging from 2 to
is presented in Table 2-3. A discussion of this 10 for instances representing various degrees of
comparison follows. training inadequacy. For situations where above
average training has been implemented, the effect
2.4.1 PSF Comparison Findings of this PSF ranges from 0.5 (SPAR-H method) to
For available time, the SPAR-H method covers the 0.1 (ASEP, Table 8.3). Neither THERP nor
entire influence range accounted for by the other HEART have multipliers for situations where
methods. For example, only ASEP, THERP, and experience and training is highly positive.
the SPAR-H method assign a failure probability of
In the SPAR-H method, an absence of procedures
1 when there is inadequate time available for crew
has a pronounced effect. The base failure rate is
response. In terms of the lower bound, the SPAR-
multiplied by a factor of 50. ASEP assigns a
H method assigns a multiplier of 0.01 for instances
failure probability of 1. THERP also assigns a
where the time available is greater than 50 times
multiplier of 50. CREAM and HEART have no
the average time required to perform the task. This
explicit assignment for situations wherein
also is comparable with multipliers used by ASEP
procedures are not available. Since there are many
and THERP. CREAM allows a reduction in the
instances of personnel performing noncontrol
failure rate when additional time is available, but
room activities without procedures, the assignment
only by a factor of 0.5. In addition, CREAM
of 1 used in ASEP seemed overly severe.
assigns its weighting factor by selecting one of
Therefore, the SPAR-H method endorses the
three common performance conditions (CPC 18,
THERP guideline for performance in the absence
19, or 20).
of procedures.
Extreme stress in the SPAR-H method is assigned
For the ergonomics and human-machine interface
a multiplier of 5. This value is higher than those
category, missing or misleading indication
suggested by either HEART or CREAM and
warrants a multiplier of 50 (SPAR-H method).
precisely the same as ASEP. However, it is less
ASEP and THERP have a more severe adjustment,
than the multiplier of 25 permissible under
ranging from a factor of 100 (THERP) to complete
THERP for instances when the cognitive state of
failure (ASEP). The SPAR-H method assigns a
the crew is such that they believe themselves to be
multiplier of 10 for situations involving poor
in a life-threatening situation. In the SPAR-H
ergonomics, as does HEART and THERP (Table
method, it was determined that the majority of
20.10). CREAM limits the influence of poor
scenarios to be reviewed would represent potential
ergonomics to a multiplier of 5, and ASEP does
situations where the extent of stress experienced
not deal with it specifically.
would be less than life threatening. All five
approaches used a lower bound (i.e., multiplier of Fitness for Duty is present explicitly as a PSF only
1) to represent nominal conditions. in the SPAR-H method and HEART. Fitness for
duty was included in the SPAR-H method because
Only the SPAR-H method and CREAM
of its appearance as a factor in a number of
differentiate among nominal, moderate, and high
operating events and also based on the uncontested

17
behavioral sciences research on the negative ASP error rates for diagnosis and action tasks was
impact of illness and circadian upset, including necessary. Further discussion of base rate
sleep deprivation, on human performance. comparisons is presented in Section 3 of this
report.
Poor work processes is present as a PSF category
in the SPAR-H method, HEART, and CREAM. 2.4.3 Relationship of PSFs to HEPs
Both CREAM and the SPAR-H method assign a Underlying the SPAR-H Method
multiplier of 5 in instances where work processes
are poor. HEART has six different error producing The basic human information processing model
conditions in which poor work processes are and its relation to PSFs is presented above in this
included. The highest multiplier available to the report. The second major component in the SPAR-
analyst is 2.0. CREAM also has four different H method is the relationship of PSFs to HEPs. The
common performance conditions with which poor third component, the SPAR-H method’s approach
work processes are associated. Only CREAM and to uncertainty analysis, is presented below in this
the SPAR-H method assign a 1 for nominal section.
conditions for work processes, and both the Unlike most HRA methods, the SPAR-H method
SPAR-H method and CREAM allow for base recognizes that a number of PSFs may have both a
error probabilities to be reduced by a factor of 0.5 positive and negative effect on performance. For
for instances where work processes are established example, training is well understood to influence
as good. performance both positively (when training
emphasizes the appropriate learned responses) and
2.4.2 PSF Changes
negatively (e.g., when training is misleading or
PSF changes were driven by several absent). In other HRA methods, positive effects on
considerations. The first consideration was PSFs are typically limited to the influence of time
consonance with the other methods. SPAR-H PSF on task performance reliability. CREAM does
information tables were presented at a meeting make allowance for the positive influence of time,
with the NRC on December 2, 1998. Based on training, and work processes PSFs upon
comments from that meeting, final adjustments performance. HEART addresses mainly the
were made to the PSFs, the PSF weights, and the detrimental effects of PSFs on performance
PSF definitions. reliability.
The second consideration was a desire to achieve The SPAR-H method assumes that most PSFs
realistic values while maintaining as much of the have positive effects that should be accounted for
1994 ASP HRA values as possible. These values in the estimation of the HEP. The SPAR-H method
had been validated by application review by also assumes that these positive effects may often
inspectors, SPAR model analysts, and HRA be a reflection of the function of the negative
practitioners. Following the update to the SPAR- effects of the PSF on performance. As shown in
H method, SPAR-H method base failure rates Figure 2-2, error probability increases as the
were compared with base failure rates associated negative influence of the PSF grows. Conversely,
with the various HRA methods. A certain amount error probabilities diminish as the positive
of analyst judgment was required, since many of influence of the PSF grows until some lower
the error types in the other methods incorporated bound is reached. Note that PSFs have a
one or more PSFs. For example, the HEART error significant effect on prediction of performance
type, “Shift or restore system to a new or original reliability (see Figure 2-2). For example, an
state on a single attempt without supervision or objective measure of fitness for duty may be the
procedures,” incorporates aspects of the time (in hours) since lack of sleep, which has a
procedures PSF and the work processes PSF. In variable influence on the performance of different
instances where it was not easily possible to people. This is shown by the distributions parallel
determine a base rate from a composite rate, to the HEP axis. The SPAR-H method models the
another HEP or base rate was used. The rates uncertainty of the HEPs at each objective level of
compared favorably and no change to the 1994 a PSF as a beta function.

18
 Greater human error
probability
1.0

Stronger error
causing effect
of the PSF
Stronger performance
enhancing effect Nominal error rate
of the PSF (1.0 E-2 for diagnosis,
1.0E-3 for actions

Lower human error

probability
1E-5

Figure 2-2. Ideal mean HEP as a function of PSF influence.

Some knowledge (i.e., limited or imperfect) of the Contributory factors to uncertainty also include the
actual shape of the individual PSF distributions is appropriateness of the nominal HEP to the actual
available, which therefore are presented as situation, the completeness of our understanding of
hypothetical distributions to aid the reader in the situation, and model uncertainty.
conceptualizing the model. Composite
distributions for PSFs are assumed to be the same For simplicity, the effect of each PSF on the HEP
as that for any individual PSF used in the method. for diagnosis or action-type task used in the
However, little is known about composite SPAR-H method is assessed through
influences of PSFs. multiplication. PSF influences are treated
independently, as is the convention in HRA. For a
It is also assumed that the uncertainties associated discussion of the potential relationships among
with PSFs affect some portion of the uncertainties PSFs, see Section 2.5. As an enhancement to
of the HEP. Uncertainty of the PSF means that it is earlier versions of SPAR-H, the method now
difficult in most instances to know the objective acknowledges that there is an advantage to being
level of a particular PSF. In addition, uncertainty able to reduce the potential to double count shared
associated with interactions among the PSFs effects among PSFs. As a result, an adjustment
influences the HEP. factor is provided with worksheets to allow the

19
analyst to reduce the effects of this potential Extra time—time available is between one to two
problem. Standard HRA also assumes that error times greater than the nominal time required, and
can be appropriately modeled with a logarithmic is also greater than 30 minutes.
function. Successful human performance may be
modeled with a logarithmic function. This may not Expansive time—time available is greater than two
be the most appropriate function when these data times the nominal time required and is also greater
are transformed into failure space. than a minimum time of 30 minutes; there is an
inordinate amount of time (a day or more) to
The levels of PSFs are negatively skewed. More of diagnose the problem.
the nonnominal ratings available for analyst
selection are negative rather than positive. Insufficient information—if you do not have
Emphasis is placed on nonnominal conditions that sufficient information to choose among the other
the regulator would expect to see. SPAR-H is alternatives, assign this PSF level.
expected to be used most often in situations where
the cases under scrutiny are worse rather than The selection of 30 minutes for a minimum time
better. Even so, use of positive PSF assignment for lowering the nominal diagnosis failure rate was
can produce HEPs on the order of 1.0E-5, below made based on several factors. This time value
which there should be only slight analyst comes from our reading of THERP diagnosis
confidence. curves supplemented by the EDF MERMOS
method. In addition, the methods differ in their
2.4.4 SPAR-H Method PSF Overview and approach to diagnosis and diagnosis-related
Definitions actions. (MERMOS combines the two). This
difference is sufficient to limit a truly empirical
This section presents, in order corresponding to determination. Also, in situations where there is
the SPAR-H worksheets, general definitions for ample time but it is believed to be less than thirty
the PSFs. As noted in other areas of this report, minutes, analysts are expected to employ the
there is overlap among PSFs. “Obvious Diagnosis” category in the Complexity
factor PSF. This was designed to correct for any
2.4.4.1 Available Time over conservatisms in the diagnosis PSF levels.

Available time refers to the amount of time that an The analyst is cautioned that “diagnosis time”
operator or a crew has to diagnose and act upon an reduces the time available for “actions” and vice
abnormal event. A shortage of time can affect the versa. Since only one time is usually provided by
operator’s ability to think clearly and consider the thermo hydraulic analysis, the analyst should
alternatives. It may also affect the operator’s keep track of the changing reduction in time.
ability to perform. Multipliers differ somewhat, Action (At-power Conditions)
depending on whether the activity is a diagnosis
activity or an action. Inadequate time—P (failure) = 1.0. If the operator
cannot execute the appropriate action in the
Diagnosis (At-power Conditions)
amount of time available, no matter what s/he
does, then failure is certain.
Inadequate time—P (failure) = 1.0. If the operator
cannot diagnose the problem in the amount of time Time available is equal to the time required—
available, no matter what s/he does, then failure is there is just enough time to execute the appropriate
certain. action.
Barely adequate time—2/3 the average time Nominal time—there is some extra time above
required to diagnose the problem is available. what is minimally required to execute the
appropriate action.
Nominal time—on average, there is sufficient time
to diagnose the problem.

20
Time available ≥ 5x time required—there is an individuals. 3 When applying SPAR-H, the analyst
extra amount of time to execute the appropriate will not have the above physical measures
action (i.e., the approximate ratio of 5:1). available. Assignment of the specific stress level
will therefore involve making an interpretation
Time available ≥ 50x time required—There is an based on operations knowledge and human factors
expansive amount of time to execute the as to the expected level of stress for a particular
appropriate action (i.e., the approximate ratio of scenario or context.
50:1).
Extreme—a level of disruptive stress in which the
Insufficient information—if you do not have performance of most people will deteriorate
sufficient information to choose among the other drastically. This is likely to occur when the onset
alternatives, assign this PSF level. of the stressor is sudden and the stressing situation
persists for long periods. This level is also
The application of time available to LP/SD associated with the feeling of threat to one’s
operation is discussed in other sections of this physical well-being or to one’s self-esteem or
report. professional status, and is considered to be
qualitatively different from lesser degrees of high
2.4.4.2 Stress/Stressors stress (e.g., catastrophic failures can result in
Stress (and level of arousal) have been broadly extreme stress for operating personnel because of
defined and used to describe negative as well as the potential for radioactive release).
positive motivating forces of human performance.
High—a level of stress higher than the nominal
Stress as used in SPAR-H refers to the level of
level (e.g., multiple instruments and annunciators
undesirable conditions and circumstances that
alarm unexpectedly and at the same time; loud,
impede the operator from easily completing a task.
continuous noise impacts ability to focus attention
Stress can include mental stress, excessive
on the task; the consequences of the task represent
workload, or physical stress (such as that imposed
a threat to plant safety).
by difficult environmental factors). It includes
aspects of narrowed attentional field or muscular Nominal—the level of stress that is conducive to
tension, and can include general apprehension or good performance.
nervousness associated with the importance of an
event. Environmental factors often referred to as Insufficient information—if you do not have
stressors, such as excessive heat, noise, poor sufficient information to choose among the other
ventilation, or radiation, can induce stress in a alternatives, assign this PSF level.
person and affect the operator’s mental or physical
performance. It is important to note that the effect
of stress on performance is curvilinear—some 3
small amount of stress can enhance performance, Hans Selye, a physician from Germany who spent
and should be considered nominal, while high and much of his professional career investigating stress,
helped to define what we know about the physiological
extreme levels of stress will negatively affect
and psychological phenomenon we refer to as stress.
human performance (e.g., see Figure 2.4, p. 34). His theory includes the general adaptation syndrome
(GAS), a nonspecific response of the body to demands
Common measures of stress have included
made upon it. This response is associated with specific
galvanic skin response (GSR), heart rate (HR), measurable physical activation of various endocrine-
blood volume pulse (BVP), numerous self-report related systems. He defined three stages of GAS:
inventories, and the measurement of chemical (1) alarm reaction, similar to “fight or flight,”
markers. For example, lowered levels of s-IgA, an (2) resistance, which is a struggle to overcome, hard
immune response marker present in saliva, have work, and limited rest or sleep; and (3) exhaustion,
been linked to increased risk of ill health in which is characterized when body systems crash,
fatigue ensues, errors become prominent, and there is
increased irritability. During this stage, the body is
vulnerable to illness (colds, flu, acne), because there is
reduced immunoresponse.

21
2.4.4.3 Complexity Figure 2-3 illustrates typical contributing factors to
complexity. Identification of these complexity
Complexity refers to how difficult the task is to
factors may be found in Braarud (1998), EPRI TR-
perform in the given context. Complexity
100259 (1992), Gertman and Blackman (1994),
considers both the task and the environment in
and NUREG-1624 (2000). The SPAR-H analyst
which it is to be performed. The more difficult the
may wish to refer to these factors when evaluating
task is to perform, the greater the chance for
the complexity PSF. It is recognized that a single
human error. Similarly, the more ambiguous the
complexity factor can result in different levels of
task is, the greater the chance for human error.
influence on human-system interaction. For
Complexity also considers the mental effort
example, mental calculations required of operators
required, such as performing mental calculations,
may be slight or, given aspects of the event, may
memory requirements, understanding the
prove to be overwhelming. The same is true for
underlying model of how the system works, and
combinations of factors. Because of this,
relying on knowledge instead of training or
assignment of the specific complexity level
practice. Complexity can also refer to physical
associated with an HEP is left to the analyst to
efforts required, such as physical actions that are
determine. At the current time, there is no
difficult because of complicated patterns of
algorithm for inferring levels of influence based on
movements.
which combination of factors is selected.

Figure 2-3. Factors contributing to task complexity.

22
For analysts who wish to differentiate between There is no obvious action PSF level assignment
rule- and knowledge-based diagnosis, in most available to the analyst. Easy to perform actions
cases the former would present less complexity are encompassed in the nominal complexity rate.
and would often be associated with a positive
rating on the procedures PSF. Knowledge-based Insufficient information—if you do not have
diagnosis and decision-making will often present sufficient information to choose among the other
the operator with greater complexity and often be alternatives, assign this PSF level.
associated with more negative ratings on 2.4.4.4 Experience/Training
procedures, including incomplete or misleading
procedures or lack of procedural guidance. This PSF refers to the experience and training of
the operator(s) involved in the task. Included in
In general, a task with greater complexity requires this consideration are years of experience of the
greater skill and comprehension to successfully individual or crew, and whether or not the
complete. Multiple variables are usually involved operator/crew has been trained on the type of
in complex tasks. Concurrent diagnosis of multiple accident, the amount of time passed since training,
events and execution of multiple actions at the and the systems involved in the task and scenario.
same time is more complex than diagnosing and Another consideration is whether or not the
responding to single events. scenario is novel or unique (i.e., whether or not the
crew or individual has been involved in a similar
Highly complex—very difficult to perform. There
scenario, in either a training or an operational
is much ambiguity in what needs to be diagnosed
setting). Specific examples where training might
or executed. Many variables are involved, with
be deficient are guidance for bypassing engineered
concurrent diagnoses or actions (i.e., unfamiliar
safety functions, guidance for monitoring reactor
maintenance task requiring high skill).
conditions during reactivity changes, and guidance
Moderately complex—somewhat difficult to for monitoring plant operation during apparently
perform. There is some ambiguity in what needs to normal, stable conditions for the purpose of
be diagnosed or executed. Several variables are promoting the early detection of abnormalities.
involved, perhaps with some concurrent diagnoses
Low—less than 6 months experience and/or
or actions (i.e., evolution performed periodically
training. This level of experience/training does not
with many steps).
provide the level of knowledge and deep
Nominal—not difficult to perform. There is little understanding required to adequately perform the
ambiguity. Single or few variables are involved. required tasks; does not provide adequate practice
in those tasks; or does not expose individuals to
Obvious diagnosis—diagnosis becomes greatly various abnormal conditions.
simplified. There are times when a problem
becomes so obvious that it would be difficult for Nominal—more than 6 months experience and/or
an operator to misdiagnose it. The most common training. This level of experience/training provides
and usual reason for this is that validating and/or an adequate amount of formal schooling and
convergent information becomes available to the instruction to ensure that individuals are proficient
operator. Such information can include automatic in day-to-day operations and have been exposed to
actuation indicators or additional sensory abnormal conditions.
information, such as smells, sounds, or vibrations.
High—extensive experience; a demonstrated
When such a compelling cue is received, the
master. This level of experience/training provides
complexity of the diagnosis for the operator is
operators with extensive knowledge and practice
reduced. For example, a radiation alarm in the
in a wide range of potential scenarios. Good
secondary system, pressurized heaters, or a failure
training makes operators well prepared for
of coolant flow to the affected steam generator are
possible situations.
compelling cues. They indicate a steam generator
tube rupture (SGTR). Diagnosis is not complex at
this point; it is obvious to trained operators.

23
Insufficient information—if you do not have critical safety functions are maintained. Therefore,
sufficient information to choose among the other if either diagnostic procedures (which assist in
alternatives, assign this PSF level. determining probable cause) or symptom-oriented
procedures (which maintain critical safety
2.4.4.5 Procedures functions) are used, there is less probability that
This PSF refers to the existence and use of formal human error will lead to a negative consequence.
operating procedures for the tasks under This being said, if the symptom-based procedure is
consideration. Common problems seen in event found to be inaccurate or awkwardly constructed,
investigations for procedures include situations then the procedures PSF should be negatively
where procedures give wrong or inadequate rated.
information regarding a particular control
Insufficient information —if you do not have
sequence. Another common problem is the
sufficient information to choose among the other
ambiguity of steps. PSF levels differ somewhat,
alternatives assign this PSF level.
depending on whether the activity is a diagnosis
activity or an action. In situations where multiple Action
transitions between procedures are required to
support a task or group of tasks, SPAR-H suggests Not available—the procedure needed for a
that the analyst adjust the PSF for complexity particular task or tasks in the event is not
accordingly. If the procedures themselves are available.
problematic, i.e., inadequate, then, the HRA Incomplete—information is needed that is not
analyst should assess the procedures and contained in the procedure; sections or task
determine whether they should be assigned an instructions (or other needed information) are
“inadequate” or “poor” rating. absent.
Diagnosis
Available, but poor—a procedure is available, but
Not available—the procedure needed for a it contains wrong, inadequate, ambiguous, or other
particular task or tasks in the event is not poor information. An example is a procedure that
available. is so difficult to use, because of factors such as
formatting, that it degrades performance.
Incomplete—information is needed that is not
contained in the procedure or procedure sections; Nominal—procedures are available and enhance
sections or task instructions (or other needed performance.
information) are absent.
Insufficient information—if you do not have
Available, but poor—a procedure is available but sufficient information to choose among the other
it is difficult to use because of factors such as alternatives, assign this PSF level.
formatting problems, ambiguity, or such a lack in
consistency that it impedes performance. 2.4.4.6 Ergonomics/HMI
Ergonomics refers to the equipment, displays and
Nominal—procedures are available and enhance controls, layout, quality and quantity of
performance. information available from instrumentation, and
Diagnostic/symptom oriented—diagnostic the interaction of the operator/crew with the
procedures assist the operator/crew in correctly equipment to carry out tasks. Aspects of human
diagnosing the event. Symptom-oriented machine interaction (HMI) are included in this
procedures (sometimes called function-oriented category. The adequacy or inadequacy of
procedures) provide the means to maintain critical computer software is also included in this PSF.
safety functions. These procedures allow operators Examples of poor ergonomics may be found in
to maintain the plant in a safe condition, without panel design layout, annunciator designs, and
the need to diagnose exactly what the event is, and labeling.
what needs to be done to mitigate the event. There
will be no catastrophic result (i.e., fuel damage) if

24
When considering panel design layout, event Poor—the design of the plant negatively impacts
investigations at U.S. commercial nuclear facilities task performance (e.g., poor labeling, needed
have shown that when necessary plant indications instrumentation cannot be seen from a work
are not located in one designated place, it is station where control inputs are made, or poor
difficult for an operator to monitor all necessary computer interfaces).
indications to properly control the plant. If there is
evidence that this is the case, a negative PSF value Nominal—the design of the plant supports correct
is assigned. performance, but does not enhance performance or
make tasks easier to carry out than typically
Examples of poor annunciator designs have been expected (e.g., operators are provided useful
found where only a single acknowledge circuit for labels; the computer interface is adequate and
all alarms is available, which increases the learnable, although not easy to use).
probability that an alarm may not be recognized
before it is cleared. Another problem exists where Good—the design of the plant positively impacts
annunciators have set points for alarms that are set task performance, providing needed information
too near to the affected parameter for an operator and the ability to carry out tasks in such a way that
or crew to react and perform a mitigating action. lessens the opportunities for error (e.g., easy to
see, use, and understand computer interfaces;
Examples of poor labeling include instances where instrumentation is readable from workstation
labels are temporary, informal, or illegible. In location, with measurements provided in the
addition, multiple names may be given to the same appropriate units of measure).
piece of equipment. Ergonomics of the plant are
also called the human-machine interface (HMI) or Insufficient information - if you do not have
the human engineering aspects. Job performance sufficient information to choose among the other
aids can also be considered a special case of alternatives, assign this PSF level.
ergonomics. However, in SPAR-H, if the job
performance deficiency is related to a procedure, 2.4.4.7 Fitness for Duty
then the preferred means of evaluating the Fitness for duty refers to whether or not the
situation is to apply this information to the individual performing the task is physically and
procedures PSF, as opposed to the ergonomics mentally fit to perform the task at the time. Things
PSF. For example, if the procedure does not match that may affect fitness include fatigue, sickness,
the equipment to be used, then the equipment- drug use (legal or illegal), overconfidence,
procedure deficiency should be noted in the personal problems, and distractions. Fitness for
procedures, not the ergonomics, PSF. duty includes factors associated with individuals,
but not related to training, experience, or stress.
During LP/SD, certain information is assumed for
the nominal ergonomics case. For BWRs this Unfit—the individual is unable to carry out the
includes availability of RCS level instrumentation required tasks, due to illness or other physical or
and RHR system instrumentation. For PWRs, this mental incapacitation (e.g., having an
includes the availability of RHR system incapacitating stroke).
instrumentation, the availability of RCS
temperature instrumentation, and the availability Degraded fitness—the individual is able to carry
of RCS level instrumentation. out the tasks, although performance is negatively
affected. Mental and physical performance can be
Missing/Misleading—the required instrumentation affected if an individual is ill, such as having a
fails to support diagnosis or postdiagnosis fever. Individuals can also exhibit degraded
behavior, or the instrumentation is inaccurate (i.e., performance if they are inappropriately
misleading). Required information is not available overconfident in their abilities to perform. Other
from any source (e.g., instrumentation is so examples of degraded fitness include experiencing
unreliable that operators ignore the instrument, fatigue from long duty hours; taking cold medicine
even if it is registering correctly at the time). that leaves the individual drowsy and nonalert; or

25
being distracted by personal bad news (such as the appropriate negative level for work practices
news of a terminal illness diagnosis of a loved PSF is assigned.
one).
Additionally, any evidence obtained during the
Nominal—the individual is able to carry out tasks; review of an operating event indicating inter-group
no known performance degradation is observed. conflict and decisiveness (e.g., between
engineering and operations), or an uncoordinated
Insufficient information—if you do not have approach to safety, is evaluated in SPAR-H as a
sufficient information to choose among the other work process problem. Schisms between operators
alternatives, assign this PSF level. and management are also considered work process
problems.
2.4.4.8 Work Processes
Work processes refer to aspects of doing work, SPAR-H does directly acknowledge potential
including inter-organizational, safety culture, work problems between the regulator and licensee as it
planning, communication, and management may affect operator and crew performance. It is
support and policies. How work is planned, assumed that problems in communication or
communicated, and executed can affect individual adherence to enforcement actions or notices are
and crew performance. If planning and indicative of work process problems.
communication are poor, then individuals may not
fully understand the work requirements. Work Finally, inadequacies in the utility corrective
processes include consideration of coordination, action program (CAP), such as failure to prioritize,
command, and control. Work processes also failure to implement, failure to respond to industry
include any management, organizational, or notices, or failure to perform root cause as
supervisory factors that may affect performance. required by regulation, is considered in SPAR-H
Examples seen in event investigations are as a work process variable. Because there are so
problems due to information not being many potential areas of concern within the work
communicated during shift turnover, as well as process category that can be assigned to a potential
communication with maintenance crews and PSF level, the analyst is directed to provide as
auxiliary operators. Measures could include much information as possible in the worksheet
amount of rework, risk worth of items in utility space provided, listing the reasons for assigning a
corrective action program backlog, enforcement particular work process PSF level.
actions, turnover, performance efficiencies, etc.
Poor—performance is negatively affected by the
The shift supervisor also plays a major role in work processes at the plant (e.g., shift turnover
work processes. Instances where the shift does not include adequate communication about
supervisor gets too involved in the specifics of the ongoing maintenance activities; poor command
event—in contrast to maintaining a position of and control by supervisor(s); performance
leadership in the control room—would indicate a expectations are not made clear).
breakdown in work processes.
Nominal—performance is not significantly
Conditions with effects adverse to quality are also affected by work processes at the plant, or work
included in the work practices category, as are processes do not appear to play an important role
problems associated with a safety-conscious work (e.g., crew performance is adequate; information is
environment. This includes retaliation by available, but not necessarily proactively
management against allegations as it pertains to communicated).
the failure event under investigation. For example,
the analyst must decide whether utility Good—work processes employed at the plant
management actions against maintenance staff enhance performance and lead to a more
have any bearing on a particular control room successful outcome than would be the case if work
action or maintenance action under evaluation. If processes were not well implemented and
the analyst believes there is such evidence, then supportive (e.g., good communication; well-

26
understood and supportive policies; cohesive For purposes of the analysis, we assumed an
crew). analyst with little or no training in SPAR-H
beyond the worksheets, and this report as a
Insufficient information—if you do not have reference. Additional assumptions appertained.
sufficient information to choose among the other For example, we assumed that the combination of
alternatives, assign this PSF level. “unfit for duty” and “inadequate time” appear with
a probability of p = 3.0E-5, corresponding to 4
2.5 Application of Multiple PSFs standard deviations above or below the mean for a
normally distributed variable. We assumed
Reer (in OECD NEA 1998) and others have noted weights present for making the ratings to be
that the direct application of multiplicative models normally distributed; however, less is known about
employing PSFs for the purpose of calculating the training and experience PSF. Therefore,
human error probabilities is only approximately probabilities from a uniform random probability
correct. There are a number of issues. These distribution were generated accordingly. Three
include using a scalar (i.e., the PSF value) to observations of experience were treated as equally
multiply a probability, the uncertainty associated probable—low, normal, and high. Each PSF was
with PSFs and the skills of analysts in making PSF treated as a discrete variable, and a value under the
assignments, and the possibility, given multiple area of the normal distribution was assigned for
negative PSFs, that the resulting conditional HEP each level of the PSF. The large number of
is greater than 1. Solving issues related to the samples used in the simulation allows for
mathematical correctness of multiplying a scalar assuming normalcy and central limit theorem
times a probability or accounting for the varying applicability (i.e., as sample size increases, the
degrees of uncertainty associated with individual sum approximates a normal distribution, and their
PSFs is beyond the scope of this document. Proper product approximates a log normal distribution).
implementation of the odds ratio (accounting for
the number of successes as well as failures) can Although the constraints discussed above were
make the determination of conditional HEPs less operative, exercise of the simulation employing
conservative and can help to reduce the propensity the adjustment factor resulted in 12% of the HEP
for negatively influenced situations to result in estimates yielding a value of 1. This is in
probabilities greater than 1. comparison to similar analyses without the
adjustment factor, employing the same sampling
We offer a simple modification of the nominal and distribution assumptions where the HEPs with
error probability that meets mathematical a value of 1 were 27% of the sample.
requirements. It is not formulated on the basis of
underlying theory regarding the relative For purposes of the simulation, the rule for
orthogonality or nonorthogonality of the eight applying the adjustment factor was the same as
PSFs used in SPAR-H. that used on the worksheets. The adjustment
factor was used when three or more PSFs were
The formula applicable to the adjustment of the assigned negative ratings. Thus, in situations
nominal human error probability is: where a strong negative context is present, we feel
that it is possible to obtain sufficiently high HEPs.
NHEP ⋅ PSFcomposite
HEP = +1
NHEP ⋅ (PSFcomposite − 1)
The type of HEP selected for this analysis was a
joint HEP, with both diagnosis and action
where NHEP is the nominal HEP. NHEP equals considered. In SPAR-H, this type of failure event
0.01 for diagnosis, and NHEP equals 0.001 for typically produces a higher HEP than diagnosis or
action. action alone and was assumed to be a good case
for examining the potential of the adjustment
Monte Carlo trials were conducted to gauge the factor to reduce the more extreme HEP values.
adjustment factor’s effect on SPAR-H results. Further analysis could be directed at review of the
Using MATHLAB, a million sample points were effect of the adjustment factor upon individual
used to generate the distribution and uncertainty. diagnosis and action HEPs. In addition, research

27
could be directed to evaluate the effect of changes assignment. The same formula may be applied to
to the dynamic ranges of PSFs upon HEPs. For the calculation of the HEP for the SPAR-H Action
example, in cases where the PSF multiplier is 50, worksheet.
Monte Carlo simulation could be used to
determine the effect in the underlying HEP The adjustment factor for Example 1 is as follows:
distribution if that range were reduced from 50 to .01 ⋅ 400
HEP = = 0.81 .
20 or 30 times the nominal HEP. .01 ⋅ (400 − 1) + 1
2.5.1 Calculating the Composite PSF
The adjustment factor is also applicable in
The composite PSF is calculated as the product of situations determined where the positive influence
the analysts ratings of all PSFs contained on the of PSFs is present.
SPAR-H worksheet. The ratings are multiplied by
one another, regardless of whether the PSF 2.5.1.2 Example 2: Action for Sequence at
influence is positive or negative. The adjustment Plant A
factor covers a large variety of combinations
Operators are required to take an action to return a
where the combined PSF influence would, if no
pump to service. The AIT reports that work
adjustments were applied, result in a probability
processes at the plant regarding the operation and
greater than one.
maintenance of this equipment are strongly
positive, the training and experience of the crew is
2.5.1.1 Example 1: Diagnosis for
high (they have received offsite training and all
Sequence at Plant A
have 10-15 years operations experience), and there
A diagnosis activity is required. Review of the is ample time (longer than 30 minutes and more
operating event by the Augmented Inspection than 2 times the time required). For this scenario,
Team (AIT) revealed the following PSF no other HEPs were mentioned in the report and
parameters to have influenced the crew’s diagnosis insufficient information was noted on the SPAR-H
of loss of inventory: procedures were misleading; worksheet. The HEP with the adjustment factor
displays were not updated in accordance with applied would be calculated in the following
requirements (ergonomics and human-machine manner:
interface); and the event was complex due to the
PSF Status Factor
existence of multiple simultaneous faults in other
parts of the plant. The report mentioned no Work processes Positive (0.8)
unusual or negative PSF influences contributing to Training Positive (0.8)
the event beyond those listed above. Time available Extra time (0.1)
The assignment of the PSF levels and associated PSF composite score = (0.8⋅0.8⋅0.1) = 0.064
multipliers is as follows (Appendices A and B may
be consulted for a list of other PSF levels): Ordinarily, application of the multiplicative
approach without application of an adjustment
PSF Status Factor factor would have yielded a human error
Procedures misleading (10) probability estimate of HEP = 6.4E-5.
Ergonomics poor (20)
Complexity moderately (2) The conditional HEP representing the composite
complex PSF influence is determined by:
PSF composite score = (10⋅20⋅2) = 400 0.001 ⋅ 0.064 0.000064
HEP = =
0.001 ⋅ (0.064 − 1) + 1 0.999064
Ordinarily, application of the multiplicative
approach without application of an adjustment
factor would have yielded a probability of 4.0. In = 6.41E-5.
that case, the analyst would have used HEP = 1.0, In the positive case, the influence of the
and have been overly conservative in that adjustment factor is less than in the first example,

28
where there is a pronounced effect of negative actions in that event sequence. When it is not, the
PSFs upon the nominal failure rate. Also, the dependency calculation should be omitted.
SPAR-H PSFs are negatively skewed, having a
larger relative range of influence for negative We believe that dependency of one task upon
situational influences. The most influential another arises from the knowledge or lack of
positive PSF is found in situations where there is knowledge of the performer of the second task
expansive time available for crew response and with respect to the occurrence and/or effect of the
opportunity for them to consult with others, such previous task. This dimension of knowledge cuts
as the tech support center, the licensing engineer, across the model of human performance presented
or the crew for the next shift. In general, SPAR-H in Figure 2-1. Mental models are updated to
is expected to be applied by NRC staff when coincide with experience and, therefore, are
performing event analyses where the majority of impacted by the same summary level factors or
situations reviewed are expected to be negative PSFs that are shown in Table 2-3 (available time,
rather than positive. complexity, stress and stressors, work processes,
experience and training; procedures, ergonomics
and human-machine interface, and fitness for
2.6 Dependency duty). For example, cues such as alarms,
A dependency method was developed in 1994 that indicators, chart recorders, CRT-based alarm lists,
yielded a dependency rating from zero to complete are what the operators attempt to attach to their
dependency. These levels were then matched to model of the situation. The more accurate the cues
the nomenclature in THERP. provided during training and subsequently stored
by the operator, the greater the tendency that he or
In 2003, the SPAR-H method was again updated, she will take the correct action. Prior actions and
this time to allow for analysts to acknowledge errors can act as current cues and establish
additional aspects of context when considering expectancies leading to propensities to look, or not
dependency. to look, for specific pieces of information. In other
The approach is meant to highlight those actions words, previous actions or recently experienced
or diagnoses that should be further reviewed and events create a mindset that guides decision
for which higher failure rates can be assumed. making.

This does not represent an exhaustive At the top level, if the operator has no knowledge
characterization of dependency but does bring a of a prior task, then that task has no effect.
degree of standardization and representation to the Obviously, this is meant from a cognitive
HRA process. Table 2-4 presents the dependency perspective. For example, if a pump is damaged
table that analysts use to assign a dependency out in the field, this operation can make pump re-
level. The leftmost column presents the criteria start impossible. If the operator has knowledge of
developed by the INL in 1999. The center column, the prior task, then we must consider what that
the five levels of dependency, is from knowledge could affect. For example, the
NUREG/CR-1278. The right-hand column was relationship between dependency and stress if the
developed for the current SPAR-H revision and prior task has failed will produce a higher level of
represents other descriptions developed for stress. This may influence subsequent task
application with the SPAR-H method that may aid performance. For available time, the important
the analyst in mapping tasks, PSFs, and other factor is whether excessive time required to take
aspects of context to the appropriate dependency one action leaves less time for the next, thereby
level. A brief discussion follows. Note that influencing the failure rate.
discretion is employed as to whether or not a A number of factors can operate to make a series
dependency calculation is warranted. The SPAR-H of errors dependent. Some of these include:
worksheets have a comments section where whether the crew performing the current task is the
analysts indicate whether or not the HEP in same or different than for the prior task; whether
question is influenced by preceding diagnoses or the current task is being performed in the same or
different system than the prior task; whether the

29
Table 2-4. SPAR-H Dependency Rating System.
Crew, Time, Location and Cue Assignments SPAR Dependency Level Additional Dependency Considerations and Basis for Interpretation
(SPAR-H 1999 Revision) (NUREG/CR-1278 1983) (SPAR-H 2004 Revision)
Same crew, close in time, same location, with or without Complete Lack of feedback, misleading feedback or masking of symptoms virtually ensures that preceding failure
additional cues will cause failure on this task as well. And/or
Situation mimics an often-experienced sequence, and sequence triggers a well-rehearsed, well-practiced
response. A lapse, slip, or mistake is virtually ensured. Or
Time demand, workload, or task complexity is such that failure on a preceding task ensures a lapse, slip,
or mistake on this task.
Same crew, close in time, different location, with or High Lack of feedback, misleading feedback, or masking of symptoms makes it highly likely that preceding
without additional cues failure will cause failure on this task as well. And/or
Situations mimic an often-experienced sequence; sequence triggers a well-rehearsed, well-practiced
response. A lapse, slip, or mistake is highly likely to result.
Time demand, workload, or task complexity is such that failure on a preceding task makes a lapse, slip, or
mistake on this task highly likely.
Same crew not close in time, same location, no High Lack of feedback, misleading feedback or masking of symptoms makes it highly likely that preceding
additional cues failure will cause failure on this task as well. And/or
Situations mimic an often-experienced sequence; sequence triggers a well-rehearsed, well-practiced
response. A lapse, slip, or mistake is highly likely to result.
Time demand, workload, or task complexity is such that failure on a preceding task makes a lapse, slip, or
mistake on this task highly likely.
Same crew, not close in time, same location, additional Moderate Lack of feedback, misleading feedback, or masking of symptoms makes it moderately likely that
cues preceding failure will cause failure on this task as well. And/or
Situations mimic an often-experienced sequence; sequence triggers a well-rehearsed, well-practiced
response. A lapse, slip, or mistake is moderately likely to result.
Time demand, workload, or task complexity is such that failure on a preceding task makes a lapse, slip, or
mistake on this task moderately likely
Same crew, not close in time, different location, no Moderate Same as above, except no cues.
additional cues
Same crew, not close in time, different location, Low Lack of feedback, misleading feedback, or masking of symptoms makes it somewhat likely that preceding
additional cues failure will cause failure on this task as well. And/or
Situations mimic an often-experienced sequence; sequence triggers a well-rehearsed, well-practiced
response. A lapse, slip, or mistake is somewhat likely to result.
Time demand, workload, or task complexity is such that failure on a preceding task makes a lapse, slip, or
mistake on this task somewhat likely
Different crew, close in time, same location, with or Moderate Likely that preceding failure will cause failure on this task as well. And/or
without additional cues Situations mimic an often-experienced sequence; sequence triggers a well-rehearsed, well-practiced
response. A lapse, slip, or mistake is moderately likely to result.
Time demand, workload, or task complexity is such that failure on a preceding task makes a lapse, slip, or
mistake on this task moderately likely
Different crew, not close in time, same location, no Low Same as above
additional cues

30
current task is being performed in a different 2.7 Uncertainty Analysis
location than the prior task; whether or not the
current task is being performed close in time to the
Suggestions For Using
prior task; and whether or not there are additional SPAR-H
cues available for the performer of the current task
2.7.1 Overview
that may serve to influence reaction time, failure
rates, and recovery. The SPAR-H method produces a simple best-
estimate HEP for use in plant risk models. The
We considered the following variables: crew
application of PSF multipliers in the SPAR-H
(same or different), time, location, and cues to
method follows a “threshold approach,” wherein
construct a dependency matrix. These four
discrete multipliers are used that are associated
parameters were combined into 16 dependency
with various PSF levels. Since these are
rules, yielding dependency ratings from low to
thresholds, the multipliers do not convey
complete dependence. A seventeenth dependency
information regarding the uncertainty associated
rule equal to zero dependency is also included in
with the multiplier. For example, a multiplier of 10
Part IV of the SPAR-H worksheets. Earlier
from the available time PSF does not represent a
versions of the SPAR-H worksheets featured a
range of multipliers (e.g., from 8 to 12). Instead,
compact version of the dependency matrix. The
the multiplier represents a shift in the nominal
full 16-level dependency matrix features
HEP. Subsequent research efforts may wish to
considerable redundancy, e.g., events by different
address the uncertainty associated with the
crews at different times will always have low
assignment of thresholds.
dependency, regardless of location or cues. While
earlier versions of SPAR-H merged redundant The eight PSFs undoubtedly contain some overlap
pathways into eight dependency rules, the current and are thus nonorthogonal. However, the SPAR-
revision of SPAR-H incorporates the full matrix of H method treats these influencing factors
16 combinations plus zero dependency. Inclusion independently. Historically, in quantifying HEPs,
of the full dependency matrix affords the SPAR–H HRA practitioners have treated these influencing
analyst greater traceability in his or her decisions factors as independent. In reality, dependence is
about dependency relationships than was unknown when simultaneously considering such a
previously possible. large group of factors (PSFs). It is unknown how
this interrelationship affects the underlying
The dependency levels match the nomenclature
probability distribution. However, a complex
used in THERP. Modification factors used in the
relationship is currently presumed. The relative
SPAR-H method were taken from the THERP
relationship (i.e., correlation) of these factors to
Tables. The approach was designed to be practical
one another is discussed separately in this section
and at an appropriate level of detail for use in a
of the report.
screening analysis.
In defining multipliers associated with mean
The right-hand column of Table 2-4 reflects
threshold times, a potentially large spectrum of
combinations of factors derived from HRA
diagnosis types is reflected. The average time for
evaluations of operating events and the application
diagnosis can, of course, vary as a function of
of HRA methods, such as ATHEANA (NUREG-
plant conditions, PSFs, and other contributions to
1624 2000), that we judged to relate to different
context. It is those factors that are used by the
levels of dependency in THERP and the SPAR-H
analysis team in determining their best estimate of
method.
the required diagnosis time, and the time available
to the crew (usually based upon thermal hydraulic
calculations). A number of assumptions
underlying human performance in conjunction
with plant performance are incorporated in the
SPAR-H method and are presented below.

31
2.7.1.1 Assumptions • It uses a noninformative prior distribution as
a starting point for the Bayesian distribution
• There is a nominal rate associated with
transformation.
diagnosis and action-type tasks. This is
consistent with traditional HRA approaches. • It preserves the overall mean value (after
multiplication of the PSFs on the nominal
• The nominal rate can be influenced by a
HEP), which is the focus of the worksheet.
number of factors determined by review of
the psychological literature. These factors are • It does not require extra uncertainty
the PSFs. Eight such factors are used in the parameter information, such as a standard
SPAR-H method. deviation or upper and lower bounds.
• For noninitiators the probability associated • It can produce small values at the lower end
with the HEP ranges from 0 to 1. The SPAR- of the distribution (e.g., <1E-6), but the upper
H method has not been designed to work with end of the distribution more properly
initiators; rather, analysts should identify represents the expected error probability.
frequencies to be used for those applications. Note that it is the upper end of the
distribution that dominates the overall
• The SPAR-H method assumes that the best
uncertainty results.
(i.e., the most informative) piece of
information available regarding the human An artifact of the SPAR-H worksheet is the
error probability is the mean. When situation where, if the majority of PSFs are
multiplying the base failure by the PSF positive, the mean values can be less than
(multiple PSFs), the resultant value is a mean 1.0E-5. In this situation, a cutoff value of 1.0E-5 is
value with its own range of uncertainty. suggested. For diagnosis tasks, the base rate
(mean) is approximated as 1.0E-2. This estimate is
2.7.1.2 Caveats based on our review of the literature of HRA
Some HRA approaches, such as THERP and methods. For action tasks, the mean nominal value
ASEP, made use of lognormal error factors, which is assumed to be 1.0E-3.
often produced upper bounds for HEPs that were HRA and human factors have not been able to
greater than one. Practitioners were aware of this demonstrate sensitivity between situations where
illogical conclusion and accepted it because of failure rates may be in the 1E-5 versus the 1E-6
base assumptions regarding lognormal range. Therefore, 1E-5 is a justifiable lower cut-off
distributions of human performance and inabilities range.
to move easily away from these normal and
lognormal distributions as a basis for these human Recent versions of the ASME Standard on PRA
performance models. (2002) have suggested that analysts consider the
maximum entropy formulation when calculating
The SPAR-H method does not use error factors, uncertainty. This approach is similar to that used
nor does it assume the use of a lognormal by the SPAR-H method, wherein we use the CNI
probability distribution. The SPAR-H method distribution (which is a special type of maximum
ultimately employs a beta distribution, which can entropy distribution). Therefore, the following is
mimic normal and lognormal distributions (in presented as a proposal as to how uncertainty may
addition to other types of distributions). be calculated using our approach. As a matter of
convenience, we assume that analysts will have
A so-called “constrained non-informative prior” access to SAPHIRE (Smith et al. 2000) software
(CNI) distribution (Atwood 1996) is used, due to when performing this calculation, but availability
several factors: of this specific software is not necessary.
• It takes on the form of a beta distribution for 2.7.2 Human Performance Distributions
probability-type events.
Basic research in human performance has
identified a number of models specific to human

32
performance. Associated with these models are is measured in bits. This reflects the fact that
distributions of human performance and people do not linearly consider each alternative in
distributions of human error. The most the order presented when making a simple
fundamental of these models are presented below. decision. Instead, they use a hierarchical process
that classifies the alternatives into the most likely
2.7.2.1 Fitts’ Law ones first. For nearly equally probable alternatives:
Research by Fitts and Seeger (1953) is seminal H = log2(n+1) .
work in the psychological literature examining
choice reaction time and is related to the SPAR-H When the choices carry different amounts of
action tasks. Fitts found that movement time (MT) information, and/or have a different probability of
was equal to the log of two times the distance from occurrence, the relationship among choices and
the starting point to the target, divided by the size reaction time is still logarithmic, and can be
of the target. The distance over size function is modeled as:
regarded as the index of difficulty (I) of that
1
movement. The distribution was determined over H i = log 2 .
hundreds of measures of hundreds of subjects. The pi
equation follows:
This formula allows H to increase as the
⎛ D⎞ probability of the event i decreases. Since we use
MT = log 2 ⎜ 2 ⎟ time as a PSF to influence HEPs, and taking
⎝ W⎠
longer time to diagnose either due to multiple
where D is the distance from the starting point of choices or to different amounts of information
the motion to the center of the target, and W is the available increases the time required to diagnose,
width of the target. this work may have relevance for the SPAR-H
method application.
Fitts’ Law demonstrates that the time required to
complete a task is an inverse function of the 2.7.2.3 Stevens’ Power Law
precision or accuracy required. The greater
Stevens’ (1951) Power Law may have relevance
accuracy or precision required in a task, the longer
for the SPAR-H method PSF for training and
the amount of time that is required for that task.
experience. A variation of this law simply states
Similarly, short time intervals decrease the level of
that the logarithm of the reaction time for a
accuracy that can be expected. In other words,
particular task decreases linearly with the
there is a speed-accuracy tradeoff. While Fitts’
logarithm of the number of practice trials taken.
Law is not currently a prescriptive formula for
Qualitatively, the law simply says only that
human error, it provides insights into time and
practice improves performance. The law has
accuracy considerations when modeling HEPs.
proven applicability to a wide variety of different
2.7.2.2 Hicks’ Law human behaviors—immediate-response tasks,
motor-perceptual tasks, recall tests, text editing,
Hicks’ (1952) Law regarding decision times
and more high-level, deliberate tasks such as game
represents research from the 1940s and 1950s
playing. Therefore, this law is applicable to the
refers to subjects’ performance when presented
SPAR-H method diagnosis and action-type tasks.
with simple choice-type tasks, and is related to the
SPAR-H method diagnosis task type. The decision Steven’s Power Law suggests that if actions are
times (T) associated with selecting a choice practiced over a period of time, performance tends
increase according to the number of binary to improve. The result of his work is a power
choices: function for performance. This could be modeled
as the inverse, which would be a logarithmic
T = IC H function. The function is:
where Ic = 150[0-157]msec/bit and H is the amount RT = aN −b
of information required to make the decisions and

33
where a = RT on trial 1, b can be approximated by
0.4, and N is the number of trails.
Optimal level

2.7.2.4 Other Laws Increasing

Efficiency of memory
Increasing
alertness emotional
Additional laws for performance are associated arousal

with these fundamental laws. For example

Meyer’s Law (Meyer et al. 1988) refers to rapid
motions, which may occur in human-computer Point of waking
interaction. This corresponds to one aspect of the
SPAR-H method PSF ergonomics and human-
machine interface:
Low High
Level of arousal or stress

D
T = A+ B
W Figure 2-4. Arousal effect on memory.
where T is the time to move to a target, D is the Arousal level can be thought of as the available
distance to the target, W is the width of the target, capacity for work. A certain amount of arousal is a
A is approximately -13 msec, and B is motivator toward learning. Too much or too little
approximately 108 msec. change can prevent learning or memory from
Meyer’s Law is a refinement of Fitts’ Law for forming. A mid-level of arousal provides the
predicting the time it takes for rapid aimed optimal level for the formation and retrieval of
movements, such as hitting a button on the screen memories. There are optimal levels of arousal for
by moving a mouse to it. (A and B are constants each task to be learned. The optimal level of
that may vary with the input device.) This model arousal is lower for difficult or intellectual
suggests that this aspect of human performance (cognitive) tasks (the operators need to concentrate
can be modeled with a logarithmic function. on the material) but higher for tasks requiring
endurance and persistence (the operators need
Meyer’s Law is derived from a stochastic more motivation).
optimized-submovement model. This model says
that movements consist of a primary submovement This arousal and performance relationship has
and a possible corrective secondary submovement significance for the SPAR-H method PSF fitness
toward a target. Meyer's Law can be used to make for duty, since that PSF covers circadian upset and
predictions of how much time it will take for a fatigue effects. It also relates to the PSFs for
user to accomplish a task involving selection of complexity and stress and stressors. As a quadratic
targets on the screen (such as icons, menus, or function, it is more properly modeled as a beta
hypertext links). than a logarithmic distribution. For a discussion of
the appropriateness of using beta in uncertainty
Another fundamental law that has driven analysis, see Section 2.6.3
significant research in human performance is
Yerkes-Dodson’s Law (1908). Yerkes-Dodson’s A final example of the available models of human
Law states that performance is an inverted U- performance that do not necessarily require the
shaped function of attention (see Figure 2-4). In assumption of a normal distribution of error is the
other words, performance is a quadratic function feature model presented by Nairne (1990). In fact,
of arousal. a variety of distributions were observed in our
review of the performance literature. The
transformations of these various distributions to a
beta distribution is legitimate and preferred over
ad hoc, log normally based techniques.

34
The feature model may have relevance for − d ij
s (i, j ) = c .
diagnosis or action tasks and is performance-
based. Representations of items in memory are
“The probability that a particular secondary
vectors that code the features of an item using a
memory trace, SMj, will be sampled as a potential
binary system, allowing features to assume the
recall response for a particular degraded memory
values of +1 or –1. Features are pattern elements,
vector, PMi, is then given by
which can be semantic or perceptual and may be
coded according to a specific sensory modality, or
modality independent coding information that can
wij s (i, j )
Ps ( SM j | PM i ) = N
be conveyed equally by one or more modalities.
Thus, a series of alarms or other indication can ∑w
k =1
ik s (i, k )
come to signify a particular plant event such as a
loss-of-coolant inventory, etc. The goodness of the where wij and wik are possible response bias
fit between the alarms and signals received by the weights.” In a practical sense, the analyst would
crew to what they have been trained to expect review an event, determine whether cues for which
determines whether or not a correct diagnosis or operators were trained on for that event type are
correct action will be undertaken. present, determine whether new/extraneous cues
yielding alternate states are present, and then
Cues can degrade if they are not stored properly or determine the match between expectancies and the
if something else is experienced that overwrites true state.
the presence of that cue. As Beaman (2000, p. 38)
notes, memory “consists of finding the best match In this model, the degree of fit with experience and
to a degraded cue amongst a set of undegraded expectancy, rather than time available, is key. This
feature vectors that reside in secondary memory model explains and is congruent with most
(SM).” In other words, based on the cues present memory phenomena. As shown by the similarity
at the time of recall, the operator attempts to function above, this model of cognitive
rebuild the original item from the available cues to performance is based on the natural log, and
reach the appropriate diagnosis. This retrieval and therefore can be modeled within the beta
reconstruction process can be modeled as follows: distribution.
“The [difference] between the degraded item and 2.7.3 Work Shift Effects
its undegraded secondary memory representation
is calculated by summing the number of The relationship of human performance to work
mismatched features, M, and dividing by the total shift is well documented. This factor can show up
number of compared features, N, as described in two SPAR-H method PSFs—fitness for duty as
[here]: it relates to circadian and fatigue effects and
ergonomics for the extent to which it reflects error
bk M k tolerant design, based on knowledge of workshift
dij = a ∑ .
effects. Dorel (1996) reviewed the importance of
N
the temporal dimension in influencing human
“The value Mk is the number of times feature performance. Even within correct performance,
position xjk does not equal feature position xjk. The there is a high amplitude effect; tasks are
parameter a is a scaling parameter that is assumed performed differently within tolerances, depending
to correspond to the overall level of attention, and on the time of day. Work processes may be carried
bk is used to weight particular comparisons if the out differently, and supervision may also be
task makes them more important than other susceptible to these influences.
comparisons. Distance, d, is then used to calculate
Even for the layperson, reduced concentration is
the similarity between the degraded vector and the
acknowledged for different times of the day. Dorel
undegraded secondary memory representation
(1996) reviewed the nuclear power plant archives
according to:
in France for the period 1981 through 1989,
assembling data for dates and times during which

35
human failures occurred. Data covered periods of Bn
both at power and LP/SD, and were collapsed Lni =
Di
across task type, location, and the number and type
of operators involved. Shift rotation factors were where Bn is the level of ability of the nth person,
found to be important; most failures within and Di is the difficulty of the ith test item.
morning or afternoon shift occur during the first
part of the shift, shortly after shift changeover and The greater the ability of the test taker, the higher
next by failure at the end of the shift. Failures the Rasch performance measure. Similarly, the
across the night shift were distributed evenly. greater the item difficulty, the lower the score or
However, relatively greater errors occurred during rating. This can be criticized, because it implies
the night shift than during the afternoon or that a f(Bn) = -f(Di). Ability is the sole determinant
morning shifts. of difficulty. This may not always be the case for
real-world tasks. For example, simple tasks can be
Some 110 failures across three facilities were perceived to be difficult by able persons when
documented. There was no significant correlation there is insufficient time available. Also, in the
between facility type and the temporal variable face of poor ergonomics, capable crew members
under investigation; data were collapsed across maintain their ability even though the task has
facilities. Frequencies for failure were greatest for been made more difficult, whereas less capable
the night shift, followed by the day shift, and then crews do not maintain their ability as well. In
by the afternoon shift. The difference between the either case, performance is expected to degrade.
morning and the afternoon shifts was significant (p This is the importance of PSFs.
= 0.035). The approximate frequencies were 41.25 Various refinements and applications of the Rasch
(night), 33.75 (morning), and 18.75 (afternoon), model have received attention. Linacre and Wright
respectively. The authors report that rest times and (2002) supplement the original Rasch approach by
slow versus quick alternation for shift change can suggesting an objective method to determine
also influence failure. The relative frequency of difficulty of a test. Frequencies are calculated for
these effects follows a linear progression that can the following four conditions:
be represented in a beta distribution.
Person N
2.7.4 Human Performance and
Right Wrong
Complexity Right a c
Person M
Historically, complexity was part of information Wrong b d
theory espoused by Shannon (1949) and others.
Over time, complexity has taken on many different Because the cases where person M or person N
meanings. Complexity, as used in the SPAR-H answers the same is uninformative, the only
method, considers multiple factors, such as informative contrasts come from cells b and c.
difficulty, ambiguity, occurrence of multiple Therefore, the probability of occurrence in cell b
faulted conditions, familiarity, and availability of can be expressed as:
job performance aids to reduce and cope with the
complexity, etc. The human performance literature p ni ⋅ (1 − p ni )
has defined complexity in various ways. One of (1 − p ni ) ⋅ p ni
the simpler approaches in the early 1960s by
Rasch was to define complexity as a function of where i refers to the test item, and pni indicates the
ability in the presence of difficulty. This was probability of person N on item i. Thus, 1-pni is the
assessed on an individual basis. This research was probability of failure of person N on item i. Cell c
first performed to determine the relative difficulty is similar, with the numerator and the denominator
of test items. Different raters were to rate the reversed. Through mathematical transformation,
items. Normalization across rater ability was Wright and Linacre demonstrate that the above
determined as part of the approach. Florin cites equation could be reduced to:
Rasch’s model in the following manner:

36
 p ni were collateral to the three studies: review of the
. impact of staffing level on performance (Hallbert
(1 − p ni ) et al. 2000); evaluation of the impact of alarm
system design on crew performance (O’Hara et al.
This can be further transformed into an equal
2000) and the influence of automation
interval linear scale with a logarithmic function
malfunctions on crew performance (O’Hara et al.
with the following form:
2000). A 39-item inventory was designed and
p ni administered. The researchers sought to determine
log =B −D whether a refined set of self-report items could be
(1 − p ni ) n i determined.
or Three performance measures were identified:
(1) operator activity against an ideal solution path;
e ( Bn − Di ) (2) rated performance across solution path, control
p ni = .
1 + e ( Bn − Di ) of plant, communication, and confidence as
measured by trained observers; and (3) system
Thus, the item or test (or task difficulty) only performance, as measured by general and scenario
depends on the attributes of item i, and Bn is the plant-specific parameter sets. Parameter
measure, depending only on the attributes of development was performed by experts who ran
person n, which can be called his or her ability. simulator trials to determine the ideal parameter
Review of these two approaches suggests that: (a) set. Experimental controls were established to
there may be merit in accounting for complexity minimize differences among subjects participating
from both a subjective, as well as an objective in the study and to reduce any differences obtained
measurement perspective; (b) difficulty as a function of differences among the
(complexity) may be more than the inverse of experimental scenarios. Subjective ratings of
ability; and (c) research in complexity should complexity were significantly related to operator
consider the potential influence of performance- performance (i.e., solution path, rated
shaping factors. Recent research addresses aspects performance, and system performance). For Study
of these three points. 3 (automation malfunction), significant
correlations were determined for all three
Research by Braarud (2002) reports three measures of performance, while for Study 2 (alarm
simulator experiments in process control that system design characteristics), they were only
establish a relationship between task complexity present for system performance. Highest
and the performance of control room crews. In a correlations were observed for subjective
comparison of measures employed, it was complexity and systems performance. Overall,
determined that the mental workload measures there was a moderate tendency for high
covered in the NASA task load index (TLX) complexity to be associated with reduced
workload measures inventory were accounted for performance.
by the concept of complexity. He proposes that
task complexity is characterized as task Once the distribution of the behavior of human
characteristics that make it difficult for an operator activity (e.g., diagnosis, action) is known, this
or crew to reach the desired end state. Inclusion of information may be used in the context of
difficulty as part of the definition of complexity Bayesian analysis to determine HEPs and their
relates to Rasch’s conceptualization of complexity. associated uncertainty distributions. An analogy
In all instances, however, the determination of from the hardware-portion of the PRA relates to
complexity must acknowledge ability that time-based component failures. For a component
mediates aspects of difficulty. To this end, family that has a constant rate of failure, the time
subjective measures of complexity must between failures is exponentially distributed.
accompany objective complexity measures. However, this information on the outcome (time to
failure) is combined with the assumption of a
This approach is embraced as part of the Halden Poisson process to determine a failure probability
Reactor Project work. Complexity data collected

37
and associated distribution. For the case of a in the behavioral sciences, but mostly to develop
Poisson likelihood, a gamma distribution provides inventories that can be used to assess psychosocial
a conjugate distribution. Thus, the resulting (i.e., clinically relevant) traits in individuals as an
distribution on the component failure rate would adjunct to therapy or to aid in the job selection
be gamma distributed. process. The Minnesota Multiphasic Personality
Inventory (MMPI; Hathaway and McKinley 1942)
2.7.5 The Categorization and is an example of an inventory that is used to
Orthogonality of PSFs support clinical intervention and therapy as well as
The majority of well-controlled studies involving the job selection processes at nuclear installations.
human performance research are conducted in Research has also been performed to find
such a manner as to determine the relationship of appropriate objective and subjective measures of
important variables two at a time. Examples of this PSFs, such as workload, complexity, stress and
type of research are presented above and are stressors, training, fatigue, and general personality
relevant to the use of PSFs in HRA. and social variables (Proctor and Van Zandt 1994;
There is limited research in the human Wickens and Hollands 1999). At least one recent
performance literature defining the simultaneous study, Hallbert et al. (2000), in reviewing staffing
interrelationships among groups of factors that are levels for advanced and current control rooms at
agreed upon to influence performance. Factor nuclear power plants, was able to estimate some
analysis statistical techniques have been employed degree of overlap among these measures.

Successful
HEP performance

Improve or degrades performance

Available Available
Time to Time to
Diagnose Act

Ergonomics Stress

Complexity

Experience/
Training Fitness
Procedures for
Duty

Work
Processes

Figure 2-5. Path diagram showing relationships among PSFs (solid lines denote high degree of
relationship, dashed lines denote medium degree of relationship); see Table G-1.

38
In an effort to guide analyst thinking regarding the attempt to quantify every aspect of PSF influences
issue of dependence and to help prevent the and relationships. We provide the above
analyst from double-counting influences when information to help inform the analyst in a
assigning PSF threshold values in HEP qualitative way. It is premature to suggest how to
quantification, the INL produced a table that model the interdependence of PSFs until more
assigns a qualitative ranking (low, medium, or about these relationships is known. We realize that
high) of the degree of correlation among the eight this area of research is a challenge germane to all
PSFs. The 64-cell table, presented in Appendix G HRA methods. During the course of conducting
as Table G-1, is only to be used as a guide. and documenting the analysis, the analyst always
has the obligation to note important relationships
Dependence among these factors could make the that he or she observes.
SPAR-H method-calculated HEPs either too
conservative or too optimistic. For example, when 2.7.6 The CNI Distribution
reviewing the deleterious effects of PSFs upon
performance, correlated factors will make the As mentioned, the CNI distribution used in the
resulting HEP more conservative than is the case. SPAR-H method is a special type of maximum
Conversely, when reviewing HEPs where strongly entropy distribution. Entropy, in the case of HEPs,
positive PSFs are present, it is possible that the represents the expectation on the logarithm of the
final HEP will be overly reduced. HEP distribution. As Atwood (1996) points out, if
Figure 2-5 presents a path diagram of the the HEP distribution has a finite range (which it
relationship among PSFs that was determined does, bounded between 0 and 1), then the function
based on Table G-1. The figure presents medium that maximizes entropy is a uniform distribution.
and high relationships and direct versus indirect A limitation in unconstrained noninformative
influences upon HEPs. distributions is that the mean value of a uniform 0-
to-1 distribution is 0.5. Consequently, the prior
From Table G-1 and Figure 2-5, a few preliminary distribution, having a mean of 0.5, would tend to
conclusions can be drawn. First, the relationship pull the posterior HEP distribution toward a mean
may be one-way, that is, PSFi may influence PSFj value of 0.5. It was this limitation that motivated
strongly, whereas PSFj may have little or no effect Atwood to develop the CNI distribution, where the
on PSFi. For example, available time has a strong constraint is that the prior distribution has a user-
influence on stress; however, stress has a low specified mean rather than a mean of 0.5.
effect on available time, which is often the product
of system conditions and equipment unavailability. The CNI is a single parameter distribution, which
is the mean. Once the mean HEP is known, the
Second, some PSFs share an inverse relationship. analyst may use Atwood (1996) to determine an
That is, as PSFi increases, PSFj decreases. For approximate distribution based on a beta
example, as job experience increases, workers may distribution. The beta distribution requires two
have a higher tolerance for (i.e., ability to deal parameters, α and β. Atwood (1996) supplies a
effectively with) stressful situations. The SPAR-H table of applicable α parameters (as a function of
method PSFs with the strongest degree of mean HEP). Figure 2-6 shows the numerical value
relationship are: of α as a function of the HEP. For example, using
the SPAR-H worksheet, if one determines that the
• Available Time on Stress—having less time
HEP has a value of 0.3, the value of α (from the
available increases stress on the individual
curve) is 0.42. The second parameter, β, is found
• Complexity on Time Available—i.e., the via the equation:
complexity of the situation is a major
determinant of whether or not the time α (1 − HEP )
β= .
available is perceived to be sufficient. HEP
Further research in HRA could be focused toward
In the case where the HEP is 0.3, β is found to be
developing the correlations that would be assigned
0.98. Now that both α and β are known, any
to each of the interaction cells. SPAR-H does not

39
analysis package containing a beta distribution multiplicative (e.g., multiplying a nominal HEP by
may be used to determine the uncertainty a PSF).
distribution of the HEP. For example, within
Microsoft’s EXCEL spreadsheet, the 5th percentile The approximate expected value and the variance
for the example HEP would be given by the of the overall HEP model are (to second order):
command:
ADDITIVE – HEP = HEP1 + HEP2
=BETAINV(0.05, α, β) Mean(HEP) = Mean(HEP1) + Mean(HEP2)
where the actual cell references to α and β are Var(HEP) = Var(HEP1) + Var(HEP2)
supplied in the command. Figure 2-7 plots the CNI
distribution for a variety of mean values, ranging MULTIPLICATIVE – HEP = HEP1 • PSF
from 1E-3 to 0.8, to illustrate the span of the Mean(HEP) = Mean(HEP1) • Mean(PSF)
uncertainty distribution. For users of the
SAPHIRE software, the only parameter that must Var(HEP) = Mean(PSF)2 • Var(HEP1)
be specified is the mean value, since SAPHIRE + Mean(HEP1)2 • Var(PSF)
has been programmed to automatically determine
the resulting associated beta distribution. where Mean( ) is the mean value, Var( ) is the
variance (recall that the variance is the square of
2.7.7 Combining Non-SPAR-H the standard deviation). These derivations assume
Information with SPAR-H that the individual parameters are statistically
independent.
Occasionally, combining disparate sources of HEP
information into a single HEP may be desirable. The mean (or expected value) of the HEP equation
For example, combining two THERP-based is a function of only two terms (to second order
actions, each with their associated lognormal accuracy). Additional HEPs may be included in
distribution and error factor (EF), will result in a the overall HEP, but the general form of the mean
single HEP estimate. However, this estimate must remains a summation of the individual means. The
then be recast into a format suitable for SPAR-H. variance of the HEP equation is a function of only
the variance (to second order accuracy) for the
Specifically, we would need to determine an
additive model and both the mean and the variance
overall mean value and, possibly, information
for the multiplicative model.
related to the uncertainty distribution (e.g., the
standard deviation). These approximate mean and variance equations
may be used to determine the aggregate
A common method of aggregating parameters that
distribution characteristics when combining
have uncertainty is to use the Taylor series
SPAR-H method and non-SPAR-H method
expansion. The statistical moments (mean,
information. Once the overall mean and the
variance, skewness, etc.) for the overall model are
variance are known, one may simply use the mean
calculated by expanding the model equation in a
value and select the CNI distribution, as advocated
Taylor series about the mean. What results from
in Section 2.6.6.
the expansion process is an equation for the
overall statistical moments, which is a function of Alternatively, one may use the method of
the variable moments and the partial derivative of moments to fit the approximate mean and variance
the model equation. to a beta distribution with that same mean and
variance. For a beta distribution, X, we have:
Most statistical texts address the Taylor series
expansion. Rather than presenting an inordinate α
amount of detail, only the results of the expansion Mean(X) =
process will be presented (Ang and Tang 1975). α +β
Furthermore, this report only illustrates two cases,
when two factors are (1) additive (e.g., summing α
Var(X) = .
two HEPs into a single action), or (2) (α + β + 1)(α + β )2

40
Figure 2-6. Alpha (α) as a function of mean HEP.

1.E+00

1.E-01

1.E-02

1.E-03

Mean = 8E-1
1.E-04
Mean = 5E-1
Mean = 1E-1
Mean = 1E-2
Mean = 1E-3
1.E-05

1.E-06
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Percentile (x100)

Figure 2-7. CNI distribution for the HEP.

41
While the approximate method described above is diagnosis may be corrected as a function of
adequate for many cases, analysis tools such as additional alarm information, i.e., an additional
SAPHIRE have mechanisms built in to facilitate alarm can be thought of as an increase in cue
model construction and analysis using Monte strength. In SPAR-H, the analyst has two means
Carlo simulation. For example, one could simply by which to represent these potential influences on
identify individual HEPs as basic events and then HEPs. The first is to perform more detailed
sum those events using the compound event modeling. The second, and suggested approach, is
feature within SAPHIRE. The uncertainty on the to make adjustment to the nominal HEP by
individual basic events would then automatically assigning the appropriate positive levels to the
be propagated though the model during the course appropriate subset of PSFs. In the above example,
of an uncertainty analysis. Nonetheless, the the work practices PSF (for additional personnel
method described above is a generic applicability being present), procedures PSF (if additional steps
and may be used when needed. strongly indicate to the operator that misdiagnosis
has occurred), and ergonomics (for new alarms
2.8 Recovery that will strongly shape the operator or crew sense
that misdiagnosis has occurred) can be used by the
Recovery as used in PRA, generally describes analyst to indicate that these factors are likely to
restoration and reparation acts required to change produce a situation where the nominal value for
the initial or current state of a system or diagnosis is overly conservative.
component into a position or condition needed to
accomplish a desired function for a given plant The current approach allows the analyst to account
state (ASME RA-S-2002). In the SPAR-H for as many recovery combinations or
method, restoration or reparation actions are opportunities as warranted, and suggests this
modeled in the fault tree or event tree logic consideration to be explicitly modeled in the logic
structure used by the analyst. Therefore, the structures.
burden to account for recovery within fault tree
logic structure lies with the analyst. Once the appropriate level of modeling has been
determined there are a number of questions the
This is in contrast to THERP, where review by analyst should pose regarding functional recovery.
second checkers, supervisors, or appearance of a They appear in Wakefield et al. (1990), who
second crew has a discrete value and can be used suggest the analyst ask:
to modify a nominal HEP. This is not the same
definition of recovery as system restoration or • Can the crew diagnose the need for recovery?
reparation as used above. THERP explicitly • Can it be accomplished in the time available?
accounts for error recovery in situations where
additional steps in procedures can indicate the • Can the equipment be put in functional
error and help to formulate the correct response, condition by personnel?
and where expected, required periodic control • Can the crew gain access to the equipment?
room scanning or system walkdowns can help to
indicate misdiagnosis. The nominal HEP may also • Are the required staff (with the right skills)
be reduced when a new, salient alarm is activated. available?
Implicit in this approach is the notion that errors in

42
 3. ANALYSIS
3.1 Base Rate Comparison completeness. One method, FRANCIE (Haney
2002), not mentioned previously in this report, was
Among HRA Methods, developed for NASA primarily as a qualitative
Including SPAR-H human error analysis method, dependent on
analyst characterization of a large number of PSFs.
To calibrate the SPAR-H method against other The method allows for quantification, and a
HRA methods, the base failure rates associated number of values from this source were included
with a number of contemporary HRAs were in this broad characterization of mixed base rates
compared. Table 3-1 compares error rates for available from HRA methods.
operator or crew actions. Here the SPAR-H
method base rate is toward the lower end of the The SPAR-H method base rates for diagnosis and
rates associated with other methods. The action were not changed. Since the various
difficulties of comparison due to PSF methods compared use different base rates, with
entanglement in the descriptions may be even different PSFs, with different levels of influence,
more of a problem in this comparison. Because of direct comparison of method rates is difficult.
this difficulty, the 1994 ASP validation, and the
firm belief of the analysts that the difference Table 3-3 presents diagnosis error type base rate
between the diagnosis and action base rates needs comparisons, which compare the SPAR-H method
to be maintained, no change in this base rate was diagnosis base rate to the base rates for diagnosis
made at this time. Future full-scale benchmarking in other HRA methods. For completeness, this
of HRA method against method could help to table includes the INTENT HRA method
resolve this issue but, resources may be better (Gertman et al. 1992).
directed toward HRA data collections, so that a These comparisons are still difficult, due to the
better basis for rates underlying HEPs might be differences in definition and the incorporation of
determined. PSFs into many of the descriptions (e.g., ASEP,
HEART). As in contemplating base rate
Table 3-2 compares mixed rates. That is, the table comparisons for operator actions as a function of
shows rates for error types whose descriptions HRA methods, a more robust comparison of the
partake of both diagnosis and action (or where a base rates could take place via a benchmarking
distinction cannot be made.) The difficulties in exercise.
making comparisons among these rates make this
primarily an information table, included for
Table 3-1. Action error type base rate comparison.
Base Rate
(5th – 95th
Method Error Type Description percentile bounds)
SPAR-H Action Task 0.001
HEART D. Fairly simple task performed rapidly or given scant attention 0.09
F. Restore or shift a system to original or new state following procedures, with some 0.003
checking
CREAM Tactical 0.001–0.1
ASEP Table 7-3. Screening critical action, assuming moderate stress, and no recovery. 0.05
THERP Table 20-2 Rule based actions of control room personnel after diagnosis, with 0.025
recovery. EF=10

43
Table 3-2. Mixed-task base rate comparison.
Method Error Type Description Base Rate
SPAR-H Task involving both diagnosis and action 0.011
HEART A. Totally unfamiliar, performed at speed with no real idea of likely consequences 0.55
B. Shifts or restores system to a new or original state on a single attempt, without 0.26
supervision or procedures
C. Complex task requiring high level of comprehension and skill 0.16
E. Routine, highly practiced, rapid task, involving a relatively low level of skill 0.02
G. Completely familiar, well-designed, highly practiced, routine task occurring several 0.0004
times per hour, performed to highest possible standards by a highly motivated, highly
trained and experienced person, totally aware of implications of failure, with time to
correct potential error, but without the benefit of significant job aids
H. Responds correctly to system command, even when there is an augmented or 0.00002
automated supervisory system providing accurate interpretation of system state
M. Miscellaneous task for which no description can be found (Nominal 5th to 95th 0.03
percentile data spreads were chosen on the basis of experience available suggesting
log normality)
FRANCIE 1. Procedural Omission 0.0059
(5th-95th 2. Error of Intent 0.085
percentile)
3. Selection Error 0.015
4. Awareness and Task Execution Related to Hazards/Damage 0.016
5. Cognitive Complexity or Task Complexity Related 0.033
6. Inspection/ Verification 0.097
7. Values/Units/Scales/Indicators Related 0.022
8. Maintenance/Repair Execution 0.041

Table 3-3. Diagnosis error type base rate comparison.

Method Error Type Description Base Rate
SPAR-H Diagnosis Task 0.01
CREAM Tactical Control Mode 0.001–0.1
Opportunistic Control Mode 0.01–0.5
ASEP Table 7-2. Screening diagnosis, assumed to be under moderate stress, given 30 0.01
minutes. EF=10.
THERP Table 20.1 Screening diagnosis. EF=10. 0.01
HEART Miscellaneous task category “M”, no description in other tasks (A-H) fits diagnosis 0.03
tasking as well.
INTENT Misdiagnose given like symptoms. Capture sequence based on stimuli. 0.057
Competing goal states lead to wrong conclusion. 0.048
Symptoms noticed, but wrong interpretation. 0.026

44
It is important to note that the SPAR-H method presented in earlier sections of this report, and are
diagnosis base rate of 0.01 is within the same present in other HRA methods.
general range encompassed by the rates for each of
the other methods. Given this and the difficulty of As an additional check on the validity of selecting
comparison, and the fact that the base rate had these PSFs, the INL reviewed operating event
some initial validation in the 1994 SPAR-H analyses present in the NRC Human Performance
method, no change in the diagnosis base rate was Event Data Base (HPED) and attempted to identify
made. instances in which effects of the SPAR-H method
PSFs could be defined. The HPED contains human
As a result of the PSF comparison detailed in factors analysis and identification of PSFs culled
Section 3, as well as additional reviews conducted, from review of licensee event reports (LERs) and
the following PSFs, presented in Table 3-4, require augmented inspection team (AIT) reports. Also
assessment as part of the SPAR-H method (2004 included in the HPED is information about the
revision) quantification process. Note that the data source, general event information, description
number of levels associated within a particular of contributing factors, personnel and personnel
PSF is, for the most part, the same for action or response during the event, and coding of factors
diagnostic tasks. The exception is “complexity,” such as communication and command and control
where, in the case of diagnosis, complex scenarios influences. A total of 196 records are contained in
were more challenging for analysts to evaluate, the database; 66 of those are from AIT sources; 42
and the addition of a level was found to allow are from LERs that correspond to ASP events. In
them greater flexibility and confidence in their instances where information was too incomplete
estimate(s). for assigning different PSF levels, the authors
reviewed the original LER or AIT sources.
Section 3.2 presents general definitions used to
assist analysts’ evaluation of PSFs. Applications to Thirty-six event summaries were coded against the
operating events demonstrating the assignment of eight SPAR-H method PSFs and their associated
PSFs are contained in Appendix F of this report. A levels of influence. Most levels of influence were
brief discussion of the occurrence of SPAR-H present in one or more of the events contained in
method PSFs in events follows here. the HPED database. Others were simply not
reported. For example, LERs typically would not
Review of PSFs selection against operating events. report that nominal time was available, or that
The PSFs in the SPAR-H method are addressed in expansive time was available. This does not mean
the behavioral sciences literature, and fit with the that these levels of available time PSF do not
information processing model of human behavior
Table 3-4. SPAR-H PSFs used in quantifying HEPs.
PSF Diagnosis Levels Action Levels Range of Influence
Available time 5 5 .01 to failure
Stress/Stressors 3 3 1 to 5
Complexity 3(4)* 3 0.1 to 5
Experience/Training 3 3 0.5 to 3
Procedures 4 4 0.5 to 50
Ergonomics/HMI 4 4 0.5 to 50
Fitness for Duty 3 3 1 to failure
Work Processes 3 3 0.8 to 5
*The number in parentheses = the number of levels associated with LP/SD

45
occur, but rather, on average, PSF information used in SPAR-H were required. An additional
tends to be reported when negative. Another level of influence for procedures was identified,
example of this is stress. Moderate or normal and the definitions for PSF levels associated with
stress is not explicitly called out in reports. The available time were revised. Some of the more
negative influence associated with high stress in important differences noted during this review are
personnel performance tends to be addressed. For presented in Table 3-5.
example, the report for the Zion Nuclear Station
(U.S. NRC 1997) shutdown event identified that a As a result of this review, changes to two PSF
large number of people in the vicinity of the influence categories, procedures and time
control room, combined with high ambient noise, available, were implemented. No basis for change
and concurrent attempts at pump restoration all in definition or range of effects for the other PSFs
contributed to a high level of stress. This stress, in was identified.
turn, contributed to improper reactivity control via After the review, an assessment was made to
poorly coordinated control rod movements. This determine whether the changes suggested in the
analysis, which provides evidence of SPAR-H revision would result in:
method PSFs in high profile events, is presented in
Appendix F. In general, it further supports the • Different values when both sets of weights
PSFs present in the SPAR-H method. were applied to the same scenario
• Analysts finding the revised LP/SD
3.2 Comparison of PSF Weights worksheets easier to apply to LP/SD
for Low Power Versus At- conditions
power • A greater face validity and corresponding
Internationally, groups such as COOPRA have analyst confidence when compared to the
sought to simplify, and by sharing information, HRA worksheets developed for at-power.
help to standardize approaches/activities
supporting LP/SD PRA. LP/SD and at-power An analysis team consisting of an HRA specialist
conditions are generally recognized to be different and two operations specialists conducted an
(more variable in the case of low power, for application to an LP/SD scenario. HEPs were
example), although they are potentially as risk calculated, and comparisons were made when
significant as many of the operations performed at applying the two types of worksheets. Based on
power (Gertman et al. 1996). For example, direct field test findings, the assignment of an additional
fuel damage and or interruption of decay heat can level of procedures PSF was made to both the at-
be considered as initiating events for LP/SD power and the LP/SD worksheets. Also, based on
conditions. Heavy load drops can have a human these findings, actions for both worksheets used
initiator, as can erroneous draining of an operation the same multiples of available time (i.e., greater
loop. 4 Currently, the NRC SPAR program has than five times nominal and greater than 50 times
been developing LP/SD models to support the nominal) to assign the appropriate PSF level of
analysis of operating events influence. Therefore, the worksheets used in the
evaluation differed primarily in one category, the
Therefore, reviews were conducted to characterize, way that time available was expressed for
evaluate the potential influence of LP/SD upon diagnosis. Diagnostic tasks are more time sensitive
human performance, and determine whether and the range and level of effect are therefore
different PSFs were required to characterize different between action and diagnosis. For
LP/SD, or if PSFs with a different range of example, the amount of time available during
influence or different weights than those currently LP/SD for activities, including diagnosis, varies
widely and generally may be less uniform than the
4
SPAR-H has been used to support NRC screening typical response time associated with conditions
analysis of human-induced heavy load drop, fuel guided by emergency operating procedures (EOPs)
misleading, and other significant actions for LP/SD during at-power.
conditions

46
Table 3-5 Assumed differences among LP/SD conditions and at-power mode.
At-power Mode LP/SD Conditions Comments
Well trained to, well defined Different initiating events may apply. For example, there may be some
initiating events may apply. procedures induced loss of coolant
accidents possible during LP/SD
conditions.
More safety systems available. Different safety systems available.
Refueling and mid loop operations Refueling operations, mid-loop operations, Level control at mid-loop operations is of
not a concern. and draindown are different than at-power concern.
evolutions and are performed less
frequently.
Transients are consistent in nature Transients are less consistent; operators in
and operators more practiced in control room and others do not practice
their response. simulator training for LP/SD activities.
Stricter limits for required operable Limits are less strict, greater number of
safety systems (how many systems systems are down for maintenance and
can be down for maintenance and repair.
repair).
Lower diversity of equipment Higher diversity of equipment Keeping track of conditions much more
configurations and operability. configurations and conditions for demanding. Different operations such as
operability. spent fuel pool handling operations
important.
Hardware contributors most likely Human errors may be more likely Multiple crews, multiple activities,
contributors to initiating events. contributors to initiating events. incomplete or infrequently used
procedures
Only 1–2 train(s) of ECCS allowed Only 2 trains are required to be operable Varies from plant to plant, set forth in
to be inoperable—as many as 6 or (4 allowed to be inoperable). Technical Specifications.
7 (counting ADS) allowed to be
inoperable.
Fewer work activities performed. Greater amount of work activities being Greater complexity may be present
performed such as tests, maintenance, and during LP/SD operations.
repairs.
Expected equipment configurations Different equipment configurations are Less frequently performed operations.
the norm. often times the norm.
Breached containment not allowed. Breached containment allowed under Restrictions may conflict with other
certain restrictions. desired shutdown evolutions, such as fuel
movement.
Predictable workload during Variable, perhaps unexpected, workload
normal at-power conditions. shifts during normally occurring shutdown
conditions.
Most activities are formally Many of the procedures being followed Example, leak in section of PCS
practiced and are heavily consist of work orders, are more custom, sampling system, but no procedure for
proceduralized. are more diverse, and in many cases have every mile of pipe and elbows exists.
not been tested. Procedure must specify order of opening
Use of mock-ups and practice of major and closing valves to isolate before
activities, especially in radiation areas is welders can come in. All testing and
often performed. Not as clear for non- equipment lineups, including what
radiation areas. systems must be in place to conduct tests,
etc. This will be specific to the area being
evaluated. Installation of temporary
bypasses or modifications is specified in
the work order. None of this will be
highly practiced, compared to startup and
shutdown procedures on which operators
are tested and trained.

47
Often, tasks may not be fully proceduralized 3.3 Approach to LP/SD
during LP/SD or, in some instances, only partially
complete procedures are available. Since this may
Comparison
be the norm, and workers often can complete their Table 3-5 suggested a number of differences
assignments without undue high error rates, the between LP/SD and at-power. It was assumed that
SPAR-H method approach to procedures was re- this might go beyond conceptualizing a generally
examined. As a result, an additional level for the different set of conditions and lead to separate
procedures PSF was generated for LP/SD. SPAR-H method worksheets, with possibly
Formerly, if procedures did not exist, the SPAR-H different levels of PSFs or different PSF ranges.
method would assign a multiplier of 50 to the base
failure rate. Another level corresponding to limited LP/SD HRA worksheets were developed and are
or incomplete procedures with a multiplier of 20 presented in Appendix B.
has been added to give HRA analysts additional
In order to determine whether there was a
flexibility to more accurately determine HEP
difference in HEP results when analysts used the
estimates. Also, the influence and or availability of
LP/SD HRA worksheets (i.e., PSF weights versus
procedures as conceptualized in THERP and other
using the at-power worksheets), a comparison of
methods did not fully include characterization of
the two types of worksheets was made. This
LP/SD situations, or other situations where skill of
comparison was performed as follows:
the craft is such that it routinely overcomes the
effects of limited or partial procedures. Making it 1. Operations and human factors analysts
easier for analysts to determine the range of separately reviewed an LP/SD event sequence
influence of procedures effect was so successful and then applied the set of PSF weights
that this additional level for procedures effect was (referred to as Weight Set A) used when
also adopted for use in the at-power HRA calculating HEPs for at-power operations.
worksheets.
2. They subsequently performed the same
The information contained in Table 3-5 suggests review and applied the forms revised for
how to redefine or renormalize the PSFs when LP/SD scenarios, which included LP/SD-
evaluating the departure of conditions from the specific weights (Weight Set B).
nominal case. That is, evaluation of deviation for
LP/SD PSFs should be conducted against what is 3. The analysts then determined the differences
expected for LP/SD conditions and not necessarily between the two resulting sets of HEPs.
for what would represent nominal conditions for These results are shown below in Table 3.6.
at-power. This is the primary reason two sets of
the SPAR-H method worksheets were developed. Weights. Using the at-power PSF weights resulted
Limited field-testing at NASA and interviews with in three out of nine HEPs (33%) associated with
human factor analysts also indicated that for the loss of inventory (LOI) scenario receiving a
normal operations in other domains, the inclusion higher failure rate when compared with HEPs
of an additional level of procedures made determined with LP/SD weighting factors. The
assignment of PSF weights less difficult for analysis team indicated that they were more
analysts. As a result, an additional level was added comfortable with the LP/SD weights, and that the
to the procedures PSF for both the at-power and LP/SD resultant values were more consistent with
LP/SD HRA worksheets. The range of influence operating experience.
for the procedures PSF was not changed, and falls Categories. In all instances, the improvements to
within the range of influence indicated by other the categories associated with time enabled
HRA methods. analysts to assign influences that better
approximated their experience. The values
assigned were within the range of influence, as
determined by review of HRA methods. Operator
performance in the three tasks (failure to initiate

48
Table 3-6. Loss of inventory with RCS pressurized HEPs Comparison of PSF influence for PSF Weight
Sets A and B
HEP Number and Description HEPpsfa HEPpsfb Change Ratio
1. Failure to diagnose loss of inventory 0.05 0.05 1
2. Failure to Initiate RCS inventory makeup 0.005 0.0005 10 to 1
3. Failure to terminate loss of inventory 1.0 1.0 1
4. Failure to recover RHR 0.00025 0.00025 1
5. Failure to re-establish RCS flow 0.003 0.003 1
6. Failure to perform secondary cooling 0.002 0.002 10 to 1
7. Failure to force feed 0.004 0.004 1
8. Failure to perform feed and bleed 0.001 0.001 1
9. Failure to establish long term re-circulation 0.002 0.00002 10 to 1

reactor coolant system, failure to perform that determined as part of process FMEA. The
secondary cooling, and failure to establish analysts were four human factors professionals
recirculation) all benefited from the extended time who had a brief training session devoted to the
horizon available to crews. SPAR-H method during INL visits to JSC under
another project. The JSC analysts found the
Although the HEPs obtained when using different screening process to be easy to use. They stated
PSF weights are obvious, the impact of the that, while the range associated with the
differences upon plant risk cannot be determined procedures influence category was sufficient, an
without accounting for the context potential of additional level for negative influences would
these failures and subsequent changes in the make the assigned PSF weight more realistic.
conditional core damage probability (CCDP). They also reported some subjectivity in the
Other samples of tasks taken from different approach, but noted that it probably was more
scenarios may result in different reduction ratios or objective than the FMEA approach that they were
similar ratios with different impacts, depending on testing concurrently. They also reported that
the initiating event sequence. conceptualizing factors such as error type, PSFs,
recovery, and dependency caused them to gain
3.4 Additional Field Testing insights regarding performance that they were able
to convey to operations personnel.
SPAR-H method (2004 revision) At-power HRA
worksheets were subject to limited field-testing at 3.4.1 Applicability of the SPAR-H Method
the NASA Johnson Space Center (JSC). In to External Events
conjunction with implementation of human
performance assessments, including human factors While the focus of this report is the SPAR-H
process failure modes and effects analysis (HF method as specifically applied to at-power and
PFMEA), three JSC processes were selected, and a LP/SD HEP determination, the heuristics
subset of tasks from each process was subjected to described herein may apply to other situations. For
a SPAR-H method evaluation. The selected example, the SPAR-H method may be applicable
processes were the J-85 engine refurbishment task, to external events such as fire, flood, seismic, and
self-contained breathing apparatus (SCUBA) tank other special failure events such as partial failures,
refilling operations, and assembly of exercise containment impacts, and plant physical security.
equipment required for the international space
When applying the SPAR-H method to potential
station.
scenarios representative of a variety of situations,
The SPAR-H method produced HEPs whose one should consider features specific to these
likelihood followed the same general pattern of additional situations. For example, an external

49
event may occur during either at-power or LP/SD factors. Overall, we observed convergence in error
operations and may occur during a transition of factors. (We postulate that this convergence
plant operational modes. Furthermore, the PSF represents the seminal influence of THERP).
impacts for an external event scenario may vary
dramatically from one case to another (e.g., a The analysis team had postulated that additional
flooding event may have a long-term duration, uncertainty is present in LP/SD conditions. This is
while a fire event may have a very short duration). because of more direct human-system interaction,
While there is no reason to believe that base less developed procedures, changing plant
failure rates or the set of PSFs used in the SPAR-H configurations, unique combinations of unusual
method would need to be changed for such plant vulnerabilities, and the increased probability
analysis, it is possible that the PSF multipliers are of mistakes and errors of commission (see
not applicable for these additional situations due to NUREG/CR-6093).
their unique character and infrequent realization. However, even though it was possible to create
To better reflect an applicable HEP, it may be error factors consistent with other HRA methods,
necessary to investigate the driving mechanisms situations still resulted where the upper bound
within situations, such as external events. While associated with HEPs could exceed 1. The analysis
this work is outside the current scope, we believe team decided to adopt a Beta distribution, as
that it would be possible to address the discussed in Section 2.7.6. The information in this
applicability of the SPAR-H method to external section is presented for archival purposes and to
events and special failure events as a part of future inform readers regarding uncertainty as expressed
development activities. in other HRA methods.

3.5 Range of Uncertainty 3.5.2 Change of Distribution Due to

Associated with HRA Truncation
Methods The earlier versions of the SPAR-H method
truncated point estimates of HEPs at 1. However,
3.5.1 Evaluation Against Other Methods in the use of uncertainty distributions, good
The INL examined other HRA methods and the practice may dictate that, if the uncertainty results
range of uncertainties employed by those methods. in a portion of the distribution are greater than 1.0
The other methods examined [THERP, ASEP, or less than 0.0, modification of the distribution is
HEART, CREAM, CAHR, and the Surry Low appropriate. (Revision to distributions is
Power and Shutdown PRA (NUREG/CR-6144)] considered in Section 2.7.6 of this report.)
actually set forth HEPs and uncertainty bounds (as The lognormal distribution is a poor choice for
opposed to methods that call for expert judgment probabilities near 1.0 because it is skewed with a
of uncertainty bounds). Although HEART and long tail on the right. An alternative approach
CREAM set forth uncertainty bounds rather than using beta distributions is presented, and we
error factors, we converted these bounds to encourage using this method.
approximate error factors for convenience of
comparison. 3.6 Change in Time PSF
These other methods were examined for error
uncertainties in both diagnosis and action tasks— Based on reviews, a major change was made to the
the two error types that the SPAR-H method uses. approach used previously (1994) to estimate the
For example, only two HEART tasks were deemed influence of time on operator diagnosis and action.
by the analysis team to be purely action tasks, Briefly, this entails a change to the time horizon,
whereas the rest of the HEART tasks were where it was determined that for actions, minutes
determined to be a mix of diagnosis and action. were not as appropriate a measure as was a
multiple beyond the nominal time. For five times
Table 3-7 shows the error factors from the other the nominal time required, HEPs are reduced by
methods compared to the SPAR-H method error 0.1. For situations in which the time available is

50
Table 3-7. Diagnosis and action error factors as a function of HRA method.
Mixed Uncertainty
Diagnosis Error Action Error Factors and Associated
Methodology Factors [HEPs] [HEPs] HEPs Comments
THERP 5(0.5), 10(0.05, 0.025) Screening Diagnosis - 5 for 10, 10 for
Screening 10(0.1, 0.01, 60 and under, 30 for 1 day
0.001) Screening actions - 10
30(0.0001)
THERP 10 (0.1, 0.01, 3(0.001 to 0.01), Larger EFs are used for HEPs smaller
0.001) 5(0.003 to 0.5) than 0.001 to reflect the greater
30 (0.0001, 10(0.0005 to 0.005) uncertainties associated with
0.00001) infrequently occurring events.
Nominal diagnosis - 10 for 30 and
under - 30 for 60 and above
3 for skipping a step, for recalling
oral instruction, reading and
recording, check reading, 10 for using
procedures in abnormal operating
condition
ASEP Post 5(0.5) 5(0.01, 0.05, 0.25)
Screening 10(0.1, 0.01,
0.001) to 30
(0.0001)
ASEP Post 10(0.1, 0.01, 5(0.02, 0.05, 0.25,
0.001) 0.2, 0.5)
30(0.0001, 10(0.001)
0.00001)
CREAM Mostly 10 (one 3 Mostly 3 (one 10 for Approximate, since actually given as
for higher HEP) lower HEP) lower bound (LB) and upper bound
(UB), expert judgment uncertainty
bounds given for specific cognitive
function failures given in Table 9,
Chapter 9. Unspecified "established
data sources" for proceduralized
behaviors such as observation and
execution, mostly expert judgment
for interpretation and planning.
Error factors from 1.3 to 10
Expert judgments in Fujita and
Hollnagel (2002)
HEART 1.5 to 3 1.5 to 45 (Very Approximate, since actually given as
asymmetric for LB and UB data based 5th and 95th
very low HEPs) percentiles are defined for the generic
tasks and used in the normal HEART
calculation to produce bounds
CAHR Similar to THERP Similar to THERP Per Personal Communication from O.
Sträter. Applied Bayesean update to
data from incident data base and then
transferred into HEPs using Rasch
model. Coincidentally, these values
are in the same range as THERP.
NUREG/CR 20 ( x > 0.000001; 5(x < 0.001) For Action (A), Recovery (R), and
6144 < 0.0001) 3(0.001 < x < 0.1) Diagnosis (D) events, uncertainty
(0.1< x) follows a lognormal distribution.
SPAR-H 10 3 Approach changed in 2003.
SPAR LP/SD 10 (0.001>x) 5 Approach changed in 2003.
20 (x<0.001)

51
50x the nominal time required, HEPs are time can be applied to address the influence of
multiplied by a factor ranging from 0.1 to 0.01. time for different situations surrounding two basic
Selection of the multiplier (e.g., 0.01) is up to the events associated with LP/SD. The other two
analyst’s discretion. This is true for both LP/SD examples address the influence of time upon
and at-power scenarios. diagnosis during an LP/SD event.
For diagnosis, time is considered differently. For Time Advantage. Having three times the amount of
at-power, extra time is assigned a multiplier of 0.1, time it normally takes the operators to place the
expansive time (i.e., greater than 24 hours) is system in service gives the operators more time to
assigned a multiplier of 0.01. For LP/SD recover from their own errors, to troubleshoot,
conditions, extra time is defined as less than or realign misalignments, and communicate with
equal to two times nominal; expansive time is others outside the control room, such as auxiliary
defined as greater than two times nominal. This equipment operators that may be required to
reflects: (a) the analyst’s greater uncertainty in perform local manipulations, and, during
assessing time available during LP/SD, and (b) emergencies, personnel staffing, the Technical
LP/SD situations where, unlike at-power Support Center (TSC), and Emergency Operations
scenarios, time has such a wide range that it is Center (EOC).
more logical to speak of multiples of this time than
to try to assign a single estimate. In estimating the Further Assumption(s). It is further assumed that
influence of expansive time, analysts should make infrequently performed, difficult diagnoses (of a
use of structured expert estimation methods such problem) will benefit more greatly from additional
as those referenced in ATHEANA (NUREG-1624 time than will routinely performed diagnoses.
2000). The new structure for this PSF for Complex, infrequently performed actions such as
diagnosis and actions is listed in Table 3-8. unique evolutions may also benefit from additional
time. However, previous HRA approaches have
By eliminating the assessment of specific times, narrowed the definition of time for diagnosis by
which is difficult for LP/SD situations, the use of minutes. This may correspond to expected
analysts are allowed to estimate a range rather than crew performance under at-power conditions. Use
a specific value. Table 3-9 presents hypothetical of a nominal estimate, given the context, appears
examples of how using the range for expansive to be more meaningful. In general, additional time

Table 3-8. Available time PSF influence for LP/SD.

LP/SD Inadequate time P(failure) = 1.0 Notes:
Available Barely adequate time 10 * Analyst's choice,
Time for (approximately 2/3 x nominal) depending on complexity
Diagnosis Nominal time 1 of diagnosis, including
Extra time 0.1 multiple factors such as
(between 1 and 2 x nominal) available help and
Expansive time (>2 x nominal) 0.1 to 0.01* likelihood of additional
cues.
LP/SD Inadequate time P(failure) = 1.0
Available Time available is approximately 10 ** Analyst's choice,
Time for equal to time required depending on complexity,
Action Nominal time 1 PPE, work environment,
Time available is > 5x the time 0.1 and ease of checking and
required recovery.
Time available is > 50x the time 0.01**
required

52
Table 3-9. Influence of expansive time on base failure rates.*
HEP
Influence of reduction
Item Event Description Complicating Conditions Expansive Time factor**
1 Diagnose Loss of Small leak with concurrent High 0.01
Inventory loss of 125 volt
instrumentation power
complicates diagnosis.
2 Diagnose Loss of Small leak, no adverse or Very low No
Inventory complicating conditions. reduction,
Nominal
3 Establish Residual Loss of bus leading to loss of High (Additional time allows 0.01
Heat Removal power to shutdown cooling. for jumpering of leads, etc)
Recovery
4 Establish bleed Loss of bus leading to loss of Low (Bleed can be 0. 1
power to shutdown cooling. performed across multiple
systems, single bus assumed
not to cut across a great
number of these systems)
* As applied to potential basic events for a pressurized water reactor (PWR) LP/SD Scenario, loss of inventory (LOI) with
RCS pressurized.
** Value multiplied against the base error rate for available time PSF.

may be expected to be beneficial in complicated to shutdown cooling. The advantage of expansive

shutdown situations involving diagnosis. time is that it allows for additional activities such
as jumpering of leads inside of cabinets,
The four events in Table 3-9 represent basic events coordination between auxiliary operators and the
that could occur in various LP/SD sequences. In control room, etc.
general, increased pressure will result in increased
loss of inventory over situations where the reactor In the final example, the crew must establish bleed
coolant system is not pressurized. Two events after LOI, with the additional complication
correspond to diagnosis of this loss of inventory presented in basic event 3, loss of bus leading to
(LOI) sequence. In the first instance, a small leak loss of power to shutdown cooling. In this
with concurrent loss of 125-V instrumentation is situation, the influence of expansive time is
postulated. The concurrent instrumentation fault thought to be relatively low. Bleed can be
increases complexity for the crew and complicates performed across multiple systems, and a single
the diagnosis. Expansive time in this situation is bus failure will most likely not affect a great
likely to allow for the crew to achieve the proper number of these systems.
diagnosis.
Additional Discussion. Most PSFs were easily
In the second diagnosis basic event, there are no assigned across domains (i.e., LP/SD and at-
complications, and in most situations the crew will power). In general, bad procedures are bad
have ample time to make the correct diagnosis. In procedures, high stress is high stress, and their
this instance having expansive time will not influence can be assessed for any domain of
appreciably change the outcome. interest. The context may influence whether or not
a certain level of stress has been reached and the
In the third basic event, as part of response to LOI, extent of the influence for stress or other PSFs
the crew must establish residual heat removal. remains influenced by other PSFs as well.
This is made more difficult by the occurrence of a
second fault, loss of bus leading to loss of power

53
Complexity ratings for diagnosis tasks executed
during LP/SD employ an additional level for
situations where the diagnosis is obvious. The
multiplier associated with obvious diagnosis is 0.1
and was determined after reviews with systems
analysts and license examiners at the INL.

54
 4. USING SPAR-H
4.1 Modeling Conventions and However, it may be the case, as in our example,
that recovery consists of two aspects: diagnosis of
Considerations the need to supply power to particular equipment
and the need to configure valves for a lineup and
SPAR-H does not go into detail regarding
then activate. In this instance, we have a joint HEP
conventional modeling considerations, such as
for diagnosis and action. These individual
level of decomposition, or provide guidance for
diagnosis and action elements constituting the
aggregating different types of subtasks to facilitate
joint HEP are not on the same level as the event
analysis. These items are situation and analyst
sequence. Their combined contribution to the
specific. Also, the PRA accounts for initiating
failure to restore (HEP3) is additive (OR) gate.
event frequencies, combinations of hardware
Therefore, after they have been computed
unavailability factors, and plant configuration
separately they are added together to = HEP3.
factors thought to have a large influence upon
HEP3 is then evaluated for dependency associated
plant risk. Differences in HRA estimates due to
with previous basic event(s). This estimate is, in
differences in task decomposition are thought not
turn, multiplied by the HEP1, HEP2, EQ, etc.,
to result in large changes in the effects of HEPs on
when determining cut sets. For our example
the overall risk estimate. The consistent use of the
above:
worksheets, the consistency of decomposition
within the same HRA, and thoughtful assignment RECOVERY HEP3 = (HEPREC diagnosis +
of PSF levels are of importance. However, a HEPRECaction).
number of questions posed by users necessitate
This is similar to working on a fault tree level in
further explanation of certain SPAR-H modeling
order to determine the HEP for the basic event
factors.
level.
Using the SPAR-H worksheets, basic events can To some, we may be somewhat conservative in
be determined in one of two ways. First, if the our approach by virtue of suggesting within a
action being considered is clearly an action with single HEP (not necessarily on the event level)
no significant diagnosis activity, then the action that if you fail to diagnose, then you most
page of the worksheet can be used, and the relative probably will fail to act appropriately (within time,
effect of PSFs detailed on the worksheets can be correct action, action not too long, or too strong,
used to modify the nominal failure rate. Then the etc). In other words, without diagnosis, action is
dependency formula addressing the influence of a much less likely. The action element of the joint
previous basic event (HEP1) can be taken into HEP does not act like a second checker, an extra
account. Below is a hypothetical example, where alarm, or some other type of redundancy or
the HEP2 for Action 1 is of this type. defense in depth to reduce the failed diagnosis
portion of the HEP. In the SPAR-H approach to
IE D A1 EQ R calculating the basic event that is a joint HEP, we
Initiating Diagnosis Action 1 Activate Recovery do not allow for acting without considering
Event (HEP1) (HEP2) Equip- (HEP3)
ment diagnosing. If diagnosis is not needed, it is
questionable to model it. There are situations
where operators may just follow the procedures
In performing the calculation, a straight rather than reach a diagnosis per se. In this case,
multiplicative approach is employed, where the the action has its own HEP. Also, for those
action HEP (0.001) is modified for applicable PSF situations where the analyst believes that
levels. In computing the cutsets, the analyst will procedures, training, or indication can enable the
convert the initiating event frequency into crew to recover from a poor diagnosis, at least two
probability and then multiply by the diagnosis modeling options exist: (1) the action portion of
HEP x the Action HEP x the calculated failure rate the joint HEP can be adjusted on the basis of these
for equipment activation x recovery by the crew. positive PSFs, or (2) the diagnosis and action

55
elements can be modeled separately on the basic expert elicitation process for determining those
event level. conditional likelihood estimates. ATHEANA
therefore offers a detailed process for uncovering
4.2 At-power errors of commission and associated error-
producing conditions.
4.2.1 Prerequisites
The ATHEANA method makes use of HFEs that
Before using the SPAR-H method, the analyst
represent actions or decisions represented in the
needs considerable knowledge of the tasks and
PRA event tree, and unsafe acts that are modeled
contexts to be rated. To this end, part of the
in fault trees. Multiple unsafe acts can result in
ATHEANA (NUREG-1624) HRA process is
either similar or different HFEs. The process is
described below, as an example of a structured
iterative in nature and has been applied to a
method to obtain the kind of knowledge needed
number of PRA issues, such as pressurized
before the SPAR-H method may be used. The
thermal shock, steam generator tube rupture, and
process described is the ideal situation. In building
fire analysis. It has also been applied
SPAR models, resource limitations, including
retrospectively to a number of high profile events
access to utility trainers and operators, will often
at U.S. commercial nuclear power plants. The ten-
preclude complete ATHEANA-like applications.
step process is outlined briefly below. It is
ATHEANA has the following advantages: its
assumed that the team has already been assembled
search process is rigorous, it differentiates among
and trained.
unsafe acts and human failure events (HFEs), and
it acknowledges the importance of PSFs and 1. Select the issue. Define the issue, interpret
context. Unsafe acts are usually modeled in fault what needs to be done, and list objectives and
trees and support the refinement and quantification human performance concerns.
of human failure events (HFEs). Human failure
events often correspond to basic events in the PRA 2. Define the scope. Select the initiating event
model. Other HRA approaches include SHARP classes and initiating events for analysis, and
and SHARP1, IEEE 1082, and ASME RA-S-2002, set priorities on characteristics of event
all of which can provide an overview of the HRA sequences.
process. 3. Describe the base-case scenario. For a given
initiator, identify the nominal operator and
4.2.2 ATHEANA Search Process
plant behavior. Begin with operationally well-
The ATHEANA HRA offers a ten-step search defined scenario(s) and well-understood
process compatible with PRA. This approach physics. Understand the trajectories of main
deviates from most first-generation HRA parameters, which provide a basis from which
approaches (such as THERP, ASEP, HCR) in that to identify and define deviations. Information
ATHEANA explicitly requires the PRA analyst to sources include the Final Safety Analysis
identify deviations from base-case scenarios Report, parameter plots, thermal-hydraulic
normally considered in PRA. The HRA analyst analysis, procedures, and operator training
must then assess the vulnerabilities in the requirements.
operator’s knowledge base (Step 5) in concert with 4. Define HFEs and unsafe acts. Review critical
complicating factors (Step 7) for the scenarios. For functions required to mitigate the event,
nondeviation base-case scenarios, there are various identifying operator actions and decisions that
methods and data that apply. However, the HRA could degrade critical functions, and produce
process associated with ATHEANA also involves a list of key actions of concern.
estimating the error-forcing context for base case
deviations and the conditional likelihood of an 5. Identify potential vulnerabilities in operator
HFE, given that context. Since HRA data for and crew knowledge base. Identify tendencies
complex, infrequently observed or considered and informal rules, and evaluate combinations
contexts with potential to challenge operators are of information rules and emergency operating
often largely absent, ATHEANA provides an procedure for vulnerabilities. Sources of

56
 information include plant procedures, human- Given that a human action has been selected for
machine interface, and training that lead to evaluation, the analyst should complete a SPAR-H
operator rules, and available emergency Human Error Worksheet. The specific information
operating procedure. the analyst needs at this point is:
6. Search for deviation from the base-case • The SPAR-H method event tree(s) or PRA
scenario. ATHEANA has advanced a event tree containing the action, and an
nontraditional discovery process for understanding of the context in which the
determining new context(s) for operating action is taking place.
events. This is more applicable to event
reconstruction or plant-specific prospective • Whether or not the human action involves
analysis. It is less common for development diagnosis or is entirely action.
of basic plant models. As part of this process, • The available time, as defined in the SPAR-H
identify physical deviations from the base method, for the human action.
case, as well as how initiators can be
different. At this stage, the analyst identifies • The level of stress, as defined in the SPAR-H
key PSFs and associated error mechanisms, method, affecting the human performers of
develops system and support dependency the action given the context.
matrices, and reviews the potential dependent
• The complexity, as defined in the SPAR-H
effects of preinitiator human actions.
method, of the human action.
Analysts also identify operator tendency and
• The relationship of the task to proceeding
error types and match various unsafe acts and
failed tasks in terms of crew, time, location,
HFEs. The deviations from the base case,
and cues—all as defined in the SPAR-H
when defined, help to establish the error-
method.
forcing context (EFC).
• Additional PSF information corresponding to
7. Identify complicating factors and links to the PSFs used in application of the method.
various PSFs. This step includes determining
additional physical conditions, hardware The above information should enable the analyst
failures, configuration problems, to rate on the first three PSFs (Available Time,
unavailability problems, missing or Stress and Stressors, and Complexity) on the
misleading indication, and confusing plant SPAR-H Human Error Worksheet. The ratings of
conditions. In addition, this step serves, in the last five PSFs (Experience and Training,
general, to expand the definition of error Procedures, Ergonomics and HMI, Fitness for
forcing context. Duty, and Work Processes) on the SPAR-H
Human Error Worksheet should be marked on the
8. Evaluate recovery factors. The analyst nominal rating. This is because these five PSFs are
completes EFC and HFE definitions by event- or plant-, or even personnel-specific and
considering the opportunities for recovering thus should not be considered at other than a
from initial errors. nominal level for the SPAR-H method base model,
9. Quantify. which is applied across events, plants, and
personnel. The opposite is true when using this
10. Document. method in event analysis. In that case, plant
operating experience, NRC notices, enforcement
4.2.3 Using the SPAR-H Method for a actions, root cause corrective efforts, etc., can be
SPAR Base Model used as forms of evidence when assigning a value
to these five PSFs. This is explained in detail in
The analyst creating a SPAR base model for a
the following sections
plant uses the SPAR-H method to assign failure
probabilities to human actions or diagnosis that Typically, dependency is refined as a part of plant-
correspond to (or are contained by) basic events in specific analysis. This is because plant specifics
the SPAR-H method event trees for the plant. may increase or lessen levels of dependency as a

57
result of equipment choice, configuration, or work Worksheet. A new SPAR-H Human Error
practices. Worksheet should be completed if any changes are
warranted. Practically speaking, few event reports
4.2.4 Using the SPAR-H Method for SPAR contain the kind of detail that will allow an analyst
Event Analysis to make extensive changes from the base model.
The starting point for using the SPAR-H method However, these reports do provide understanding
for event analysis is the SPAR base model of the of a degraded situation that can provide the basis
plant to be analyzed. To analyze a selected human for updating PSFs in the existing model. Much
action of interest in an event analysis, the analyst more likely, only a few specific facts will be
first refers to the SPAR-H Human Error contained in the event reports, which will provide
Worksheet completed for that human action in evidence for differences between the base model
support of the SPAR base model for the plant. The and the event.
analyst then goes through the SPAR-H Human 4.2.5 Sources of Information for Applying
Error Worksheet, point by point, deciding whether the SPAR-H Method to Events
the context of the event being analyzed requires
that changes be made to the base model analysis. The analyst can probably never have too much
Event analysis is event-specific. information on which to base SPAR-H method
ratings. Generally, the problem is just the opposite.
In addition to reviewing equipment availability Primary sources of event information are licensee
associated with the event, the analyst must now event reports, augmented inspection team reports,
consider in detail, information about the five PSFs and the NRC’s Resident Inspector. In recent
that were automatically rated nominal on the base events (such as the degraded condition of the
model’s SPAR-H Human Error Worksheet. reactor vessel head discovered by Davis Besse), a
Additional information that the analyst normally root cause analysis team report also may be
receives at this point is: available as a source of PSF information.
Table 4-1 presents some suggestions on where the
• The quality of the experience/training, as analyst may acquire the needed information for the
defined in the SPAR-H method, of the plant eight PSFs and four dependency factors used on
personnel performing the action. the SPAR-H Human Error Worksheet.
• The quality of the plant procedures, as
The nominal ratings on the SPAR-H Human Error
defined in the SPAR-H method, used in
Worksheet are intended for use where ratings
performing the action.
actually are “average.” For instances where
• The quality of the ergonomics, as defined in insufficient information is available the analyst
the SPAR-H method, of the plant controls and now assigns a new category “insufficient
displays used in performing the action. information available."
• The fitness for duty, as defined in the SPAR- Other sources of information that may be of use to
H method, of the personnel performing the the SPAR-H analyst include:
action.
• Morning report for the event
• The quality of the plant and personnel work
• Plant procedures
processes, as defined by the SPAR-H
definitions used in the performance of the • Other inspection team reports about the plant
action.
• Plant layout diagrams and control room panel
In summary, to use the SPAR-H method for diagrams or pictures
operating event analysis, the analyst examines the
• Operator exam results
changed context of a given human action and
decisions during the event and decides whether or • Plant training materials
not to change any of the eight PSFs or dependency • Event reports from other plants on similar
factors on the base model’s SPAR-H Human Error events

58
Table 4-1. PSF sources of information
for SPAR-H.
Needed Information Source(s)
Available Time Is nearly always available in both LERs and AIT reports.
Stress/Stressors Information about stress is more likely found in an AIT report than in
an LER, but in either case it will most likely require some inference
on the part of the analyst. The analyst is more likely is to find
physical and environmental stressors reported directly. Resident
Inspector (RI ) and Plant Management (PM) sources. Operator
examiners are potential information sources.
Complexity Complexity is generally inferred in LERs and in most AITs. RI and
operator examiners are potential information sources.
Experience/Training AIT report may list for an individual and may comment on
shortcomings of training programs. Less likely found in an LER.
Operator examiners are another information source.
Procedures May be able to request procedures used and evaluate then personally.
Otherwise, LERs sometimes and AITs almost always contain
procedures review.
Ergonomics/HMI Explicit only in some AIT reports where ergonomics were a concern
and a human factors expert was part of the team. However, can often
infer, even from an LER. In some instances, the analyst may be able
to acquire drawings and procedures indicating location and type of
indications used in following procedures. Expect that major
influences such as lack of level indication or partial loss of control
room annunciation following battery run down will be contained in
either event document.
Fitness for duty Only available in an AIT report; LERs almost never contain this
information. RI and PM are other sources.
Work processes May be able to infer from an LER. Deficiencies are generally detailed
explicitly in AIT reports. RI and PM are other sources.
Dependency Factor: Crew Usually in LERs and AIT reports. RI is another source.
Dependency Factor: Time Usually in LERs and AIT reports.
Dependency Factor: Location Generally in LERs and AIT reports.
Dependency Factor: Cues Much more likely to be referenced in AIT reports than in LERs.

• Inspection recommendations 1. Determine whether the event applies to at-

• Information from walkdowns. power or LP/SD. Select the appropriate
SPAR-H Worksheet. Both worksheets are
4.2.6 Completing the SPAR-H Human identical, except slightly different PSF
Error Worksheet multipliers are used to reflect different human
Figure 4-1 illustrates the steps in completing the error probabilities, depending on the mode of
SPAR-H Human Error Worksheet. The text below operation.
describes the mechanics of completing the 2. Enter header information at the top of the first
worksheet, as it applies to at-power and LP/SD for page of the appropriate SPAR-H Worksheet.
base models and events.

59
 • The name of the plant being rated (e.g., • in Section C. A negative PSF influence is
Peach Bottom 2). one with a multiplier weighting greater
than 1.
• The name of the particular SPAR
initiating event being rated (e.g., Loss Of • Record the final diagnosis HEP. This
Off-Site Power). value is equal to the value from Section B
or C, depending on whether or not the
• The basic event code for the basic event
adjustment factor was applied.
being rated.
5. Complete Part II, Action. If action is not
• The name or initials of the Event Coder. involved, skip Step 5 and proceed with Step 6.
• The context of the basic event being rated • Enter header information from the first
(e.g., the previous basic events in this page (see Step 2 above) at the top of the
particular sequence on the SPAR event third page of the SPAR-H Worksheet.
tree).
• Rate the eight PSFs for the action portion
• The description of the basic event being of the basic event human action by
rated (e.g., operator fails to throttle high- marking one checkbox for each PSF. Note
pressure injection to reduce pressure). that when a rating is made, the rater
3. Decide whether the basic event human action should document the reason for the rating
involves diagnosis and, if so, mark the proper in the block to the right. Each PSF allows
checkbox. Use the “Why?” line below the the analyst to indicate if there is
checkboxes to describe why diagnosis is or is insufficient information to rate it. PSFs
not involved. If diagnosis is not involved, skip with insufficient information are given the
Step 4 and proceed with Step 5. same multiplier as nominal PSFs.
4. Complete Part I, Diagnosis. • Enter header information from the first
page (see Step 1 above) at the top of the
• Rate the eight PSFs for the diagnosis
fourth page of the SPAR-H Worksheet.
portion of the basic event human action by
marking one checkbox for each PSF. Note • Transfer the multipliers next to the marked
that when a rating is made, the rater checkboxes to the blanks in Section B.
should document the reason for the rating Multiply the string of multipliers by 1.0E-
in the block to the right. Each PSF allows 3 to calculate the diagnosis failure
the analyst to indicate whether there is probability.
insufficient information to rate it. PSFs
• If there are three or more negative PSF
with insufficient information are given the
influences, calculate the adjustment factor
same multiplier as nominal PSFs.
in Section C. A negative PSF influence is
• Enter header information from the first one with a multiplier weighting greater
page (see Step 2 above) at the top of the than 1.
second page of the SPAR-H Worksheet.
• Record the final action HEP. This value is
• Transfer the multipliers next to the marked equal to the value from Section B or C,
checkboxes to the blanks in Section B. depending on whether or not the
Multiply the string of multipliers by 1.0E- adjustment factor was applied.
2 to calculate the diagnosis failure
6. Complete Part III, Calculate the Task Failure
probability.
Probability Without Formal Dependence.
• If there are three or more negative PSF
• Enter header information from the first
influences, calculate the adjustment factor
page (see Step 2 above) at the top of the
fifth page of the SPAR-H Worksheet.

60
 Start

At Power or
At Power LPSD
LP/SD?

Use At Power Use LP/SD

Worksheet Worksheet

Enter Event
Description

Diagnosis,
Diagnosis Action, or Action
Both?

Both

Use Diagnosis
Use Diagnosis Use Action
and Action
Worksheet Worksheet
Worksheets

Evaluate PSFs

Calculate HEP

3+
PSFs that are
Yes
Greater than
1?

Apply HEP
No
Correction Factor

Yes Dependency?

Recalculate HEP
No
for Dependency

Final HEP

PSFs
Available Time Complexity Procedures Fitness for Duty
Stress/Stressors Experience/Training Ergonomics/HMI Work Processes

Figure 4-1. Basic flow diagram for completing the SPAR-H worksheets.

61
 • Transfer the diagnosis failure probability (low to complete). Mark the dependency
from page 1 (if diagnosis is involved) and level in the dependency condition table.
the action failure probability from page 3
• Adjust the level of dependency if a
to the blanks in Part III.
second, third, or fourth checker is being
• Calculate the task failure probability modeled as part of recovery. For example,
without formal dependence by adding the If the event is the third basic event (second
diagnosis failure probability to the action checker) in the sequence, dependency
failure probability. must be no less than moderate; if it is the
fourth event (third or fourth checker), the
7. Complete Part IV, Dependency.
dependency must be no less than high. If
• Decide whether there is a reason why there is a compelling reason for less
failure on previous basic events should not dependence, do not apply the rule, but
be considered in the rating of the present document the reason in the block above
basic event. If, for example, different the rule.
personnel are involved and have no
• Calculate the task failure probability with
knowledge of previous tasks, and there are
formal dependence by transferring the task
new cues for their tasks, this may make
failure probability without formal
some proportion of their actions
dependence from Part III into the equation
independent of these tasks. In other cases,
for the proper level of dependence.
previous tasks may increase or decrease
Clearly indicate which equation was
subsequent difficulty, and, thus, some
applied.
degree of dependency may be present. If
there is reason for not considering HRA In addition to the steps above, the analyst may
dependency quantification, then document wish to follow additional best-practice guidelines.
the reason(s) on the line at the top of Part The following suggestions are designed to
IV. improve the quality of the SPAR-H Worksheet:
• Decide whether the crew performing the • When evaluating any procedures that
basic event is the same crew that failed the contributed to an event, the specific
previous basic event in the sequence. procedures that were used by personnel should
be indicated. This information would
• Decide whether the basic event is close in typically include the number, section, and
time to the previous failed basic event in subsection for emergency, annunciator, or
the sequence. This range extends from a standard operating procedures.
few seconds to a few minutes.
• If the analyst refers to external sources of
• Decide whether the basic event takes place information such as event or plant databases,
in the same location as the previous failed plant personnel, or operations experts, these
basic event in the sequence. sources should be noted in the SPAR-H
• Decide (if not close in time) whether Worksheet.
additional cues were available following • When filling out the SPAR-H Worksheet, the
the previous failed basic event in the analyst should indicate any pertinent
sequence. These cues can be additional information available on indicators and
parameter displays, alarms, or procedures displays. This information should be noted
and procedural steps providing guidance under the Ergonomics/HMI PSF and should be
to the operator. reflected in the selection of the multiplier for
• Follow the choices on the four bullets the PSF.
above through the dependency condition • In the absence of plant-specific information,
table to arrive at a level of dependency the analyst may wish to assume a crew shift
change every 6 to 8 hours for long-duration

62
 LP/SD events. This assumption affects
dependency estimations. In SPAR-H, it is
generally assumed that the replacement crew
has the same number and personnel
complement. Any variance from this
assumption should be noted in the SPAR-H
Worksheet.
It is good practice, upon completion of the SPAR-
H Worksheet, to have a second person review the
PSF assignments and calculations. The SPAR-H
Worksheet includes a place at the bottom of each
page for an external reviewer to initial to signify
his or her approval.

63
64
 5. DISCUSSION
5.1 Differences between At- mistakes are included, only “in control room”
errors are typically modeled.
power and LP/SD
A number of significant differences between the The data collection efforts of this endeavor have
human actions, errors, and influences important to resulted in the identification of several important
at-power operations and those important to LP/SD influences on human performance during LP/SD.
operations have been identified. Through the evaluation of reports, event-based
data sources, and interviews, we identified
Aspects of the following features are identified as procedures, human engineering and HMI, training,
unique and important to LP/SD operations: the organization factors, and communications as
kinds of human interaction and events; the classes, significant contributors to human error and
modes, and types of human errors (and actions); actions. This is consistent with the set of PSFs
influences on human performance; and plant used in the SPAR-H method. Complexity was not
conditions and configurations. Unlike at-power explicitly referenced in these reports, but was
operations, all classes of human actions and errors thought to be implicitly evident by the members of
(i.e., initiator, preaccident, and recovery) seem to the analysis team. This was verified by discussions
play a significant role in LP/SD operations and with licensed operators at the INL.
events. In particular, human-initiated events
usually are not explicitly treated in at-power Procedures are important in modeling human
PRAs. It is typically assumed that human-initiated errors in full-power PRAs. In LP/SD procedures
events while at at-power can be captured in data are also important, however, work packages and
collected at the component, system, or plant level work control orders may also figure prominently
and have no detrimental impact on response in plant activities. For at power and LP/SD
following the initiator. For LP/SD events, conditions communications, complexity, and
however, human-induced initiators both inside and situation awareness should be reviewed as
outside the control room constitute a significant potential influences on human performance. The
portion of observed errors. In addition, event-based data evaluations strongly indicate that
dependencies frequently exist between the contributions from multiple influences are
activities leading to the initiating event and those common for human actions and errors during
required for recovery. These may not be apparent LP/SD and at-power events. Also, the available
at all times because there exist during LP/SD a time for event response, frequently an important
large number of potential system configurations fact in human performance at at-power, does not
and the system response can vary as a function of appear to be as critical during LP/SD. (Exceptions
the current configuration. are likely for events initiated shortly after
shutdown when decay heat is high, and for events
Data evaluations indicate that the mistakes versus that can progress unnoticed for extended periods
slips subset of commission dominates the types of time).
and modes of human errors, which occur during
LP/SD. In addition, mistakes occur both inside and In the context of nuclear power plant operations,
outside the control room during LP/SD. The more workload and stress are often closely related.
direct human-system interactions characteristic of Increased workload and stress were often cited in
LP/SD operations can result in mistakes, which in the literature as potential contributors to human
turn, lead to unwelcome consequences. In contrast, error during LP/SD. The presence of a much larger
the human errors explicitly modeled in full-power staff, including less-experienced personnel at the
PRAs are typically errors of omission [e.g., the plant, as well as the influence of extended work
NRC Generic Letter 88-20 10 CFR 50.54(g)] does periods, can play significant roles in increasing the
not require errors of commission to be modeled in workload of operators. However, plant staff
licensee individual plant examinations), and when interviews indicate that high workload and stress,
while potentially significant during LP/SD, did not

65
appear to be at detrimental levels at the plant. It following loss of cooling, for example, is
was stated that during an outage, the size of the sufficiently extended so that delays in recovering
operations crew is expanded and the shift from errors may have less impact on risk.
organization is changed to minimize the impact of
the increased workload and to reduce the stress of Despite these differences in uncertainty regarding
outage of operations. These measures were cited day-to-day operations and in personnel
by the staff as effective in minimizing the impact availability, the SPAR-H method has kept the
of outage operations on workload and stress. same PSF grouping used for full-power operations.
Therefore, we believe that the addition of Many influences, such as shift work effects, time
personnel may increase organizational load, as of day, and other personnel factors, are captured
opposed to individual load. Increased under the fitness-for-duty PSF. Reliance upon
organizational load can result in unsafe acts, work orders as opposed to more formal procedures
leading to human failure events. Perhaps future is captured under work processes. Until progress
research will evaluate staffing and organization in determining the key aspects of work process
factors more directly. factors such as safety culture upon risk can be
placed in a qualitative framework, they are best
Unlike full-power operations, LP/SD operations handled as PSFs whose effects are multiplied
are routinely performed under complicated against base failure rates.
conditions. For example, much greater emphasis is
placed on manual control actions. Also, personnel 5.2 Compliance with ASME
not normally at the plant (e.g., headquarters
engineers and contractors) and others not as
Standard on PRA
familiar with the plant’s day-to-day work practices While the current version of the SPAR-H method
and normal operating procedures may be was being completed, the standard for the conduct
performing tasks that can affect safety. In addition, of PRA (ASME RA-S-2002) was released. The
problems can exist in terms of the operators’ following indicates areas where the SPAR-H
ability to observe the state of the plant and the method is in compliance and where it is not, and if
configuration of its equipment. not, why. In general, there are some differences
between the SPAR-H approach and the ASME
To their credit, operators generally perform well in standard. These are as follows.
the face of continuously changing plant conditions
and configurations. Frequent changes in the plant 5.2.1 Organization
situation have the potential to result in changes in
ASME PRA is organized according to nine PRA
the potential consequences of events because it
elements, including initiating events (IEs),
influences the availability of backup (and, in some
accident sequence analysis, success criteria, and
cases, front-line) equipment in event responses.
systems analysis. SPAR-H begins with the
Additionally, the changing plant environment
assumption that PRA has provided detail regarding
during LP/SD calls attention to importance of
IEs to the HRA analyst. The same is assumed for
communications in supporting outage activities
information regarding accident sequence analysis,
safely and to appropriately respond to LP/SD
success criteria, and systems analysis. The HRA
events. Also, equipment operated manually on a
analyst is expected to interact with the rest of the
routine basis during LP/SD operations and the
PRA team.
response to LP/SD events are typically achieved
through manual human actions rather than 5.2.2 Documentation
automatic equipment response.
ASME PRA provides guidance regarding the
These differences from at-power operations help control of PRA documentation. SPAR-H does not
create a situation in LP/SD where certain types of address this issue. We believe that individual
errors may be more likely and their consequences entities have the responsibility to develop and
less observable. However, a significant mitigating maintain their own configuration control over
factor is that, after the first few days of an outage, PRA/HRA documentation. The only requirement
the time required for fuel to become uncovered for documentation control cited in this report

66
refers to the SPAR-H requirement that analysts 5.2.6 Probability Assignment
and reviewer sign each HRA worksheet and that
ASME lists two other important high-level
they document the assumptions behind each PSF
requirements: to use a systematic process to
assignment. Otherwise, only the documentation
estimate human error probabilities (THERP and
that is required in a credible HRA should apply.
ASEP are given as examples, p. 50), and to use
This would include maintenance of an engineering
mean values when providing HEP estimates. As
design file and copies of relevant plant procedures,
part of the development process, SPAR-H was
databases, task analysis results, or any other
reviewed against THERP and ASEP for nominal
pertinent information used by the HRA analyst to
values for diagnosis and action, and reviewed
support the HRA. Defining these requirements
again when determining the range of PSF
further is not within the scope of this report.
influences. SPAR-H provides mean values that can
5.2.3 Expert Judgment be used in conjunction with a constrained non-
informative before producing upper and lower
ASME reviews the role of expert judgment bound estimates for the HEP.
methods and presents requirements. SPAR-H
attempts to provide a simple, consistent, and easy- The ASME standard suggests (p. 48) that
to-use method that reduces expert judgment as assessment of the probabilities of the postinitiator
much as possible. We acknowledge that there are HFEs be performed using a well-defined and self-
potentially some domains such as extreme events consistent process. This process must address the
where the context is so novel or strongly negative plant-specific and scenario-specific influences on
that the applicability of the base failure rates in human performance and between human failure
SPAR-H and the range of effect for PSFs as well events within the same sequence. The SPAR-H
as the applicability of these PSFs should be method process is simple, internally consistent,
reviewed. and relatively easy to apply. Two base rates are
proposed: one for diagnosis and one for action;
5.2.4 Activity Types and the same eight PSFs are required for
ASME differentiates among preinitiator HRA, evaluation of every HEP.
postinitiating event activities, and special cases
The ASME standard suggests that the analyst
such as restoration, and provides definitions for
check the reasonableness of HEPs in light of the
logic model elements. SPAR-H assumes that the
plant’s operating history, procedures, operational
underlying model for human performance is the
practices, and experience.
same for pre- versus postinitiator actions and
diagnoses and that any differences can be The ASME standard requires that some
accounted for by the proper assignment of PSFs. assessment of the uncertainty in HEPs be
conducted. The SPAR-H method provides an
5.2.5 Work Processes
approach by which uncertainty associated with
ASME requirements call attention to identifying HEPs can be determined.
work processes that could introduce mechanisms
that simultaneously affect equipment in different 5.2.7 PSF Inclusion
trains of either redundant or diverse systems. ASME notes a number of performance influences
SPAR-H recognizes this as a significant failure that the analyst should account for. These
mode but does not present requirements regarding influences correspond to the PSFs in SPAR-H.
the search for common-cause mechanisms. These include procedures, including specification
SPAR-H suggests that if analysts consult of the procedural guidance (EOP and AOP);
approaches such as those in SHARP1 or ergonomic considerations, including cues;
ATHEANA, that they be responsive to the ASME availability of instrumentation; adequacy of
common-cause requirements. special tools and accessibility; complexity of the
required response; quality of training and
experience; the time available and time required to

67
respond; and “some measure of scenario induced The ASME standard specifically requires that the
stress” (see p. 53). documentation provided by the analyst describe
the analysis processes used and details of
Finally, ASME is specific regarding those factors underlying assumptions made. The SPAR-H
that must be present on a plant-specific basis in method requires the analyst to document on the
order to give credit for operator recovery actions. HRA worksheets assumptions made when
SPAR-H is nonspecific on this issue, and analysts assigning PSF values. We believe that analysts
should consider the guidance of ASME. should always state in the body of the PRA/HRA
5.2.8 Dependency and Procedures key assumptions regarding human performance,
availability of equipment and indications,
As is fitting with a standards approach, ASME procedures, unusual aspects of the event in terms
does not prescribe a unique approach to of plant configuration or conditions, unusual or
dependency calculation or uncertainty analysis. unexpected system response, and assumptions
SPAR-H prescribes an approach to dependency regarding the time available and time required for
determination and quantification. Suggestions for operators and crew to respond.
uncertainty assessment are provided as well.
5.2.10 Supporting Requirements for HRA
Dependency assignment in SPAR-H is adapted
from THERP. The consideration of plant and Unlike the ASME standard, SPAR-H does not
scenario-specific factors is evidenced in the provide guidance on how to develop PRA-specific
identification and characterization of errors that screening approaches. We believe it best that the
are modeled and in the analyst shaping factor level analyst apply the worksheets completely for any
assignments. Defaults for individual PSFs are for application including screening. The analysis
situations where no influence by the PSF is should be consistent and scaled to the same level
expected or where insufficient information of decomposition to allow for proper
regarding the PSF exists. Assessment of task quantification. Likewise, the categorization
dependency is straightforward and allows for scheme identified in the ASME for capability
consideration of elements of context, such as the categories I, II, and III is not called out specifically
crew depending on the same cues originating from in SPAR-H. Rather, those activities and insights
either inadequate or adequate procedures and or are used to support human response logic model
instrumentation. Quantification follows THERP (i.e., fault tree) development or as part of the PSF
guidelines. evaluation process. For example, independent
checking by a second operator can be represented
5.2.9 Procedures Review and in fault tree logic model structure or taken into
Documentation account when evaluating the quality of work
processes PSF. The appropriateness of the
ASME RA-S-2002, HLR-HR-E Section 4.5.5.1 assignment is a function of the application and the
requires analysts to perform a systematic review of scenario.
relevant procedures. The review of procedure
information, coupled with walkdowns, interviews, 5.2.11 Recovery
and review of event databases, is suggested as part
of the review of SPAR-H basics covered in The ASME standard specifies a number of
Section 1 of this report. However, beginning with considerations for recovery, self-recovery, or
first-generation HRA, analysts have reviewed recovery by other crews, and lists a number of
procedures in support of the HRA process. conditions. The SPAR-H method is much more
SPAR-H assumes that any HRA analyst will do brief and only advises that recovery be considered
the same. The range of procedures that can be in the logic structure of fault tree models used by
subject to review can include normal operating analysts, and refers them to SHARP or the ASME
procedures, abnormal operating procedures, standard for guidance. The SPAR-H method does
emergency operating procedures, and test, not go into specifics regarding credit for use of
calibration, and surveillance procedures. written checkoff lists, work shift, or daily checks
of components, etc.

68
The ASME standard specifically directs analysts technique, such as the SPAR-H method, use of
to define a set of HFEs as unavailability’s of such coarse HEP values bypasses the utility of the
functions, systems, or components at the current approach.
appropriate level of detail. The SPAR-H method
calls for the definition of HFEs in the same 5.2.14 Task Characteristics
language but recommends fault tree and event tree Task characteristics to consider that are spelled out
structure congruent with the concepts of HFE and in the ASME standard and that constitute part of
context expressed in ATHEANA and other the analysis process undertaken by the SPAR-H
second-generation methods. analyst include:
5.2.12 Timing • Number of subtasks
The ASME standard specifies the analyst • Complexity and difficulty of required actions
determine the specific timing of the accident
sequence, relevant cues and time window for • Task performed inside or outside of the
completion. The SPAR-H method assigns control room
weighting factors based on available time • Addressing both diagnosis and execution for
windows. Timing and appearance of cues are each postinitiator
noted in the ergonomics/HMI PSF. PSF details
should be noted in the comments column of the • Diagnosis—detecting, evaluating, and
worksheets. Although not required by SPAR-H, deciding response
the analyst may wish to develop functional • Execution—performing activities indicated
timelines for events to aid them in characterizing by diagnosis.
human-system performance and discrete changes
in PSFs. 5.3 NASA Guidelines
The ASME standard specifies that the definition of The SPAR-H method is also consistent with a
HFEs should coincide with accident-specific number of the elements outlined in the NASA
procedural guidance. The SPAR-H method leaves PRA Procedures Guidelines (Stamatelatos and
the determination of HFEs up to the analysis team. Dezfuli, 2002). The NASA guide reviews different
human interaction (HI) classification strategies in
5.2.13 Screening contemporary HRA. The system most widely used
employs the nomenclature HI-A,-B and -C, which
For screening analysis, the ASME standard
correspond to pre-initiating event interactions,
suggests use of conservative estimates or detailed
initiating event-related interactions, and post-
analysis of HEP estimation in dominant accident
initiating event actions, respectively.
sequences. The SPAR-H method is compatible
with this suggestion, although the SPAR-H The SPAR-H method is capable of providing
method attempts to be more realistic than a pure estimates for HI-A and HI-C, and suggests that
screening by offering a relatively large dynamic frequencies for initiating events be used from
range and mixture of PSFs for consideration in the operations or industry data where possible. The
quantification process. NASA guide and the SPAR-H method break down
HI-C (postinitiator responses) into cognitive
Earlier versions of the ASME standard specified responses and action responses. This bears a direct
values for screening. That feature has been similarity to the overarching diagnosis/action task
removed from the current standard. Many of the taxonomy used by the SPAR-H method.
values prescribed were not consistent with
approaches used in shutdown HRA, such as that The NASA guide distinguishes between skill-,
included in NUREG/CR-6144 (1995). It also rule-, and knowledge-based (SRK) behavior
called for relatively high HEPs, on the order of (Rasmussen 1979) and omission- and commission-
0.50, for postinitiator screening values. This based errors. These are not inconsistent with the
feature was also removed. For a detailed screening SPAR-H method. SPAR analysts should take the

69
knowledge domains used by operators in management factors can be important but are not
conjunction with their training into account when usually explicitly modeled in HRA. However, they
identifying errors for consideration in their can be inferred by their impact on procedures,
analysis. interface, training, and other variables.
For approaches focusing on errors of commission, The SPAR-H method considers a subset of
see Pyy (2000), Bieder et al. (1999), or Forester et organizational factors and work processes, in
al. (NUREG-1624 2000). The SPAR-H method explicit fashion, for the impact on human
uses a blended rate for errors of omission and performance and allows for quantification based
commission. This is thought to be sufficient for on this information. The SPAR-H method also
model building and for most other SPAR model directly calls out fitness for duty as an influential
applications. Just as in the case of SRK, use of variable regarding human response. Aspects of
omission and commission is not explicitly called fitness for duty are more implicitly dealt with in
out in our approach. However, it should be present the NASA guide. The NASA guide also reviews
as part of the mindset of the HRA analyst as s/he approaches to screening analysis versus detailed
considers possible errors. analysis. The SPAR-H method is already a
simplified approach, but could be used to assist in
Consistent with most HRA approaches, the NASA either qualitative or quantitative screening
guide suggests that task analysis be used in analysis. Because of the mandatory consideration
support of HRA. This is part of the SPAR-H of PSFs within the SPAR-H method, the approach
method and many other HRA approaches it uses in support of the screening analysis process
beginning with THERP. would have to be considered detailed HRA
In terms of similarities, more remarkable perhaps screening analysis. The NASA guide also refers
is the degree to which PSFs suggested by the the reader to the NUREG/CR-1278 (1983)
NASA guide and those included in the SPAR-H quantification model and five levels of
method overlap. For example, typical PSFs dependency. The SPAR-H method uses a similar
suggested by both the NASA guide and the SPAR- approach to the suggested quantification from
H method for consideration include procedures, THERP but provides supplemental qualitative
quality of human-machine interface information needed to assign the level of
(ergonomics/HMI in SPAR-H nomenclature); dependency before quantification. In terms of
training and practice (training and experience in uncertainty, the reasons for use of the beta
the SPAR-H method); task complexity distribution are summarized above in this report.
(complexity in the SPAR-H method); stress level
(stress in the SPAR-H method); time available or 5.4 General Discussion
time urgency (available time and stress in the The SPAR-H method has been developed to be
SPAR-H method); environmental conditions (part straightforward, easy to apply, and based on both a
of ergonomics/HMI in the SPAR-H method); human information-processing model of human
communication (not directly covered in the SPAR- performance and results from human performance
H method); and previous actions (covered by studies. This simple HRA approach contains a
dependency in the SPAR-H method). number of enhancements, including calibration of
its base failure rates and range of PSFs influence
The separation of environmental conditions from
against other HRA methods. This version of the
ergonomics in the NASA guide reflects the high
SPAR-H method also contains a revised approach
degree of consideration given to the effects of
to uncertainty analysis, employing a beta
microgravity on task performance. Also,
distribution that obviates problems experienced in
communication between mission specialists and
earlier versions when applying error factor
ground control operations or among crew, many of
approaches. The SPAR-H method also provides a
whom might be using English as a second
adjustment factor to reduce the likelihood of
language, emphasizes the importance of
double correcting the influence of PSFs due to
communication for that domain. The authors of the
their relationship to other PSFs.
NASA guide also note that organizational and

70
The SPAR-H method has been refined as a result RELAP-5 thermo-hydraulic computer code with
of experience gained during its use in the the IDA crew performance model.
development of over 70 SPAR plant models for
the NRC; in limited HRA applications for dry It may take time to reach consensus as a
cask, spent fuel storage; in implementation of risk- community regarding how to model and quantify
informed plant inspection notebooks; and through the effect of work processes on performance,
third-party application to other domains, such as because work processes have an indirect and
aerospace. Although there have not been formal pervasive influence on performance. The extent to
trial applications by teams of independent analysts, which work process elements, such as poor
the method has been revised based on the configuration control, work order discrepancies,
experience of teams in applying the method over the amount of re-work, infractions, risk worth of
the past ten years, as described above. The method corrective action backlog, and more objective
does not differentiate between active and latent elements, can be measured will help us to
failures; their identification and modeling is the formulate a manner for including work processes
decision of the analyst. It is thought that the same in PRA through HRA.
PSFs and base failure rates are applicable to either
type of error. The base error rates contained in the Traditionally, accounting for the influence of
worksheets for actions and diagnosis include multiple shaping factors with multiple levels of
omission and commission types of errors. The influence without imposing a high degree of expert
tendency for an omission or commission to be consensus judgment on the HRA process has
more important in contributing to an individual proven difficult. The SPAR-H method attempts to
human failure event can be modeled by the analyst make the assignment of human error probability
using subtask level of decomposition in building more reliable by presenting analysts with a
supporting fault trees. consistent set of PSFs for evaluation. The HRA
search process for determining unsafe acts, given a
Although recognition that work processes are particular context, still remains a challenging task
important is not new in HRA, the explicit for the PRA/HRA analyst, but this is the
incorporation of work processes is relatively new. information that is brought to the SPAR-H method
In instances where the effects of particular PSFs, for quantification. The need to provide sound
such as work processes are difficult to determine, qualitative assessments of PSFs is amplified as the
the range of effect used in the SPAR-H method SPAR-H method applications move from basic
reflects the treatment of the work process PSF in plant PRA model development to event analysis
other HRA methods. For example, the work and HRA analysis for specific issues.
processes’ range of effect in the SPAR-H method
is enveloped by identification of a range of effect HRA has become a central topic to PRA, in part
for work process PSF in two methods, CREAM due to the compelling notion regarding the
(Hollnagel 1998) and HEART (Williams 1988). importance of psychology, action, and mental
The range in the SPAR-H method is within the activities in everyday life. In the nearly 30 years
bounds suggested by these methods. since WASH 1400 was issued (Rasmussen 1975),
appreciation of the importance of human error in
Other recent efforts related to work process nuclear power plants has increased considerably.
analysis include that of Weil and Apostolakis Starting with a crude diagnosis model based on
(2001). Dynamic approaches to work process time, HRA practitioners now look more
analysis at nuclear facilities is presented in Shukri systematically at complexity, context, situation
and Mosleh (1998). They treat crew performance awareness, and complicating conditions as factors
factors, including aspects influenced by work in addition to time that may influence crew
processes, in conjunction with dynamic plant diagnosis and response. Theory and model
response determined by plant thermal-hydraulic building have continued with general recognition
calculations. See also Chang and Mosleh (1998) of the importance of special issues such as errors
for an overall description of the integration of of commission, cognitive control, and work

71
Table 5-1 SPAR-H method assessment.
HRA Method Criterion SPAR-H Explanation
Method Rating
The method should be Moderate Experience with use of the SPAR-H method
objective and internally suggests training time is minimal, and the level at
consistent. which analysis is conducted is consistent
throughout.
The method should High The use of defined levels of influence, base rates,
produce consistent and worksheets serves to produce consistent
results. results, given the same input.
It should possess Undetermined An existing analysis would have to be revisited at a
temporal halves later date and analysts would have to evaluate the
reliability. same HEPs, after which the extent to which the
first and second evaluations matched would have to
be determined.
It should possess face and High The method is consistent with human behavior
construct validity. modeling and appears to capture the majority of
elements considered in HRA. The relationship to
human performance at NPPs is obvious.
It should be documented Moderate The method has been field-tested for some years
and field-tested. and has been in use as part of SPAR model
development, but the documentation is only now
becoming widely available.
It should produce High The rates and range of influence have been
estimates consistent with calibrated against existing methods and offers the
the practice of HRA and analyst the range of PSFs that would make
with operating application findings consistent with operating
experience. experience
It should be applicable Moderate The method has been extended to ground
across domains. operations for aerospace with some degree of
success; application to additional domains is
needed to further establish robustness of the
method.
It should be subject to Limited The method is not yet widely distributed nor
peer review. reviewed.
Output from the method High The method was designed to produce output
should be compatible suitable for use in PRA event or fault tree logic
with existing or emerging structures.
PRA logic structures.
The method should be High The method employs predefined fields, including
easy to apply. PSFs, basic error rates, and method for dependency
assignment and quantification. Determining the
final HEP is relatively easy.
The method should be Moderate With publication in NUREG format and
easy to obtain. availability on the Web and in conference
proceedings, information about the method is
easily obtained.

72
processes. In the last 10 years, the importance of diminished capacity to perform? If so, how?
errors of cognition has been recognized. It is And if so, are serial searches or parallel searches
likely that some time in the future there will be a more susceptible to disruption? How can the
uniform treatment of uncertainty in HRA. impact of this phenomenon be reduced? One
approach is to simplify the work environment to
Human interaction with advanced technologies reduce workload. But if everything except
is a frontier for which data are needed. Task emergency situations is simplified, is the
sharing between human and intelligent systems workload reduced to the point that we are now
in robotic environments is now becoming more, rather than less, vulnerable? How are
commonplace. Much of it proceeds because the skills and alertness maintained so that they do
technology has become available. A time is not have a negative impact on safety-significant
envisioned in the future when this technology situations? What role should designing multi-
will be introduced into the control room or modal systems (vision, audition, touch) play in
perhaps balance-of-plant activities. For example, building cognitive support systems for future
consider self-maintaining, self regulating generation plants or backfits to existing plants?
systems. In fact, the importance of advancing or
extending the experimental techniques now It is apparent that HRA data collection must be
available to collect HRA data cannot be sponsored to meet the needs of the future while
overemphasized. More is probably known about applying the resources available to risk
the factors that cause crews to fail than to informing current decision-making.
succeed. For example, complexity is
acknowledged as an influence on performance. Once the answers to some of these issues are
Complexity may impact the searches that crews found, the character of HRA will be further
conduct to support hypothesis generation. Does improved.
it cause a narrowing of the search space or just a

73
74
 6. REFERENCES
Anderson, J. C., Cognitive Psychology and Its the 4th International Conference on
Implication, Fourth Edition, San Francisco: Probalistic Safety and Management, pp.
W. H. Freeman and Company, 1995. 2468–2474, New York, Springer-Verlag
Publishers, September 13–18,1998.
Ang, A., and Tang, W. H., Probability Concepts in
Engineering Planning and Design, Volume 1, Davis Besse, Root Cause Analysis Report for
New York: John Wiley & Sons, 1975, pp. Davis Besse, First Energy Corporation Utility
198–199. Root Cause Analysis Team, Oakwood, Ohio,
2002.
ASME RA-S-2002, Standard for Probabilistic
Risk Assessment for Nuclear Power Plant Dorel, M., “Human Failure in the Control of
Applications, American Society for Power Systems: Temporal Logic of
Mechanical Engineers, 2002. Occurrence and Alternating Work Times,” in
Neville Stanton (ed.) Human Factors in
Atwood, C. L., “Constrained Non-informative
Nuclear Safety, London,Taylor and Francis,
Priors in Risk Assessment,” Reliability
1996.
Engineering and System Safety, 53, 1, 1996,
pp. 37–46. EPRI TR-100259, Approach to the Analysis of
Operator Actions in Probabilistic Risk
Baddeley, A.D. Human Memory: Theory and
Assessment, Palo Alto, Electric Power
Practice, Oxford: Oxford University Press,
Research Institute, 1992.
1990.
Fitts, P. M., and Seeger, C. M., “S-R
Beaman, C. P., “Computational Explorations of
Compatibility: Spatial Characteristics of
the Irrelevant Sound Effect in Serial Short-
Stimulus and Response Codes,” Journal of
Term Memory,” Proceedings of the Twenty-
Experimental Psychology, 46, pp. 199–210,
Second Annual Meeting of the Cognitive
1953.
Science Society, August 13-15, Philadelphia,
PA, pp. 37–41, 2000. Fujita, Y., and Hollnagel, E., “From Error
Probabilities to Control Modes:
Bieder, C., LeBot, P., and Cara, F., “What Does a
Quantification of Context Effects on
MERMOS Analysis Consist In?” PSA ’99,
Performance,” Proceedings of the
American Nuclear Society, Washington, DC,
OECD/NEA/CSNI Workshop Building the
pp. 839–845, 1999.
New HRA: Strengthening the Link Between
Blackman, H. S., and Byers, J. C., ASP/SPAR-H Experience and HRA, Munich, Germany,
Methodology, Internal EG&G report for the 2002.
U.S. Nuclear Regulatory Commission, 1994.
Gertman, D. I., and Blackman, H. S. Human
Braarud, P.O. Complexity Factors and Prediction Reliability and Safety Analysis Data
of Crew Performance (HWR-521), OECD Handbook, New York, John Wiley
Halden Reactor Project, Norway, 1998. Interscience, 1994.
Braarud, P. O., “Simulator Experiments as Gertman, D. I., et al., “INTENT: A Method for
Empirical Basis for Performing Shaping Estimating Human Error Probabilities for
Factors in HRA,” in Probabilistic Safety Decision Based Errors,” Reliability
Assessment and Management (PSAM 6). Engineering & System Safety 35, pp. 127–
Bonano, E. J., Camp, A. L., Majors, M. J., & 136, 1992.
Thompson, R. A. (eds), Elsevier Science Ltd.,
Gertman, D.I., et al., “Representing Context,
2002, pp. 327–332.
Cognition, and Crew performance in a
Chang, Y. H., and Mosleh, A. A, “Dynamic PRA shutdown risk assessment,” Reliability
using ADS with RELAP5 Code as its Engineering and System Safety, 52, pp. 261-
Thermal Hydraulic Module,” Proceedings of 278, 1996.

75
Gertman, D. I., et al., Review of Findings for Medin, D. L., and Ross, B. H., Cognitive
Human Performance Contribution to Risk in Psychology, 2nd Edition, New York,
Operating Events. NUREG/CR-6753, 2002. Harcourt Brace, 1996.
Gertman, D. I., et al., Human Performance Meyer, D. E., et al., “Optimality in Human Motor
Characterization in the Reactor Oversight Performance: Ideal Control of Rapid Aimed
Process, NUREG/CR-6775, INL/Ext-01- Movements,” Psychological Review, 95, 3,
01167, 2002. pp. 340–370, 1988.
Hacker, W., Arbeitspsychologie: Psychische Nairne, J.S. “A Feature Model of Immediate
Regulation von Arbeitstätigkeiten, Berlin, Model,” Memory and Cognition, 18, pp. 251–
Verlag der Wissenschaften, 1986. 269, 1990.
Hallbert, B.B., Sebok, A., and Moriseau, D.A. Newell, A., and Simon, H. A., Human Problem
Study of Control Room Staffing Levels for Solving, New Jersey, Prentice-Hall, 1972.
Advanced Reactors, NUREG/IA-0137,
NUREG-1624, Rev. 1, Technical Basis and
November 2000.
Implementation Guidelines for a Technique
Haney, L. N., Framework for Assessing Notorious for Human Event Analysis (ATHEANA),
Contributing Influences for Error Division of Risk Analysis and Applications,
(FRANCIE): Overview and Generic User Office of Nuclear Regulatory Research, 2000.
Guidance, INL/EXT-01-01014, Idaho
NUREG/CR-1278, Handbook of Human
National Laboratory, 2002.
Reliability Analysis with Emphasis on
Hannaman, G. W., and Spurgin, A. J., Systematic Nuclear Power Plant Applications (THERP)
Human Action Reliability Procedure Final Report, Sandia National Laboratories,
(SHARP), EPRI NP-3583, Palo Alto, Electric 1983.
Power Research Institute, 1984.
NUREG/CR-4674-V17-26, Precursors to
Hathaway, S. R., and McKinley, J. C., The Potential Severe Core Damage Accidents:
Minnesota Multiphasic Personality Inventory, 1992, A Status Report, Oak Ridge National
Minneapolis, University of Minnesota Press, Laboratory, 1992.
1942.
NUREG/CR-4772, Accident Sequence Evaluation
Hick, W. E., “On the rate of gain of information,” Program (ASEP) Human Reliability Analysis
Quarterly Journal of Experimental Procedure, Sandia National Laboratories,
Psychology, 4, pp. 11–26, 1952. 1987.
Hollnagel, E., Cognitive Reliability and Error NUREG/CR-6093, An Analysis of Operational
Analysis Method (CREAM), Oxford, Elsevier, Experience During Low Power and Shutdown
1998. and a Plan for Addressing Human Reliability
Assessment Issues, Brookhaven National
IEEE Draft Standard 1574, “Best Practices for
Laboratory, 1994.
Conducting Human Reliability Analysis
(HRA),” in review, expected June 2004. NUREG/CR-6144, Surry Low Power and
Shutdown PRA, Brookhaven National
IEEE Standard 1082, “Guide for Incorporating
Laboratory, 1995.
Human Action Reliability Analysis for
Nuclear Power Generating Stations,” IEEE NUREG/CR-6618. Validation and Verification for
1997. SAPHIRE, Version 6.0 and 7.0, Idaho
National Laboratory, 2000.
Linacre, J. M., and Wright, B. D., “Understanding
Rasch Measurement: Construction of OECD NEA 98, “Critical Operator Actions –
Measures from Many-facet Data,” Journal of Human Reliability Modeling and Data
Applied Measurement, 3, pp. 486–512, 2002. Issues,” Volume 1 and Appendix F
(Questionnaire Responses), in Organization
for Economic Cooperation and Development

76
 (OECD) Nuclear Energy Agency (NEA) Final Stamatelatos, M., and Dezfuli, H., Probabilistic
Task Report, Principal Working Group No. 5, Risk Assessment Guide for NASA Managers
Task 94-1, Paris, France, February 1998. and Practitioners, Version 1, Washington,
D.C., National Aeronautic and Space
O’Hara, J.M., et al., The Effects of Alarm Display,
Administration (NASA), 2002.
Processing, and Availability on Crew
Performance, NUREG/CR-6691, 2000. Stevens, S. S., Mathematics, Measurement, and
Psychophysics, in Handbook of Experimental
Proctor, R. W., and Van Zandt, T., Human Factors
Psychology, New York: John Wiley & Sons,
in Simple and Complex Systems. Boston,
1951
Allyn and Bacon, 1994.
Straeter, O., “The CAHR Method,” in
Pyy, P., “Contribution from Finland: Framework
NEA/CSNI/R(2000)17: Errors of Commission
for Analyzing Commission Errors,” in
in Probabilistic Safety Assessment, Paris,
NEA/CSNI/ R (2000) Errors of Commission
Organization for Economic Cooperation and
in Probabilistic Safety Assessment,
Development, Nuclear Energy Agency, July
Organization for Economic Cooperation and
2000.
Development, Nuclear Energy Agency, Paris,
June 2000. US NRC Generic Letter 88-20: “Individual Plant
Evaluation for External Events for Severe
Rasmussen, J., On the Structure of Knowledge—A
Accident Vulnerabilities,” 10 CFR 50.54(f).
Morphology of Mental Models in a Man-
Washington, D.C., United States Nuclear
Machine Context, RISØM-2192, Roskilde,
Regulatory Commission, 1988.
Denmark, RISØ National Laboratory, 1979.
U.S.NRC, Licensee Event Report on Oyster Creek
Rasmussen, N., NUREG/WASH 1400, Reactor
Unit 1, Report 05000219/92-007-00,
Safety Study: An Assessment of Accident
Washington, D.C., U.S. Nuclear Regulatory
Risks in Commercial Nuclear Power Plants,
Commission, 1992.
1975.
U.S.NRC, Licensee Event Report on Dresden 3,
Rasch, G. Studies in Mathematical Psychology: I.
Report 05000249/96-004-00, Washington,
Probabilistic Models for Some Intelligence
D.C., U.S. Nuclear Regulatory Commission,
and Attainment Tests, Oxford, Nielsen and
1996.
Lydiche, 1960.
U.S.NRC, Augmented Inspection Team Report on
Sanders, M. S., and McCormick, E.J., Human
Zion Station 1, Report 50-295/97007,
Factors in Engineering and Design, Seventh
Washington, D.C., U.S. Nuclear Regulatory
Edition, New York, McGraw-Hill, 1993.
Commission, 1997.
Shallice, T. “Specific impairments of planning,”
U.S. NRC, Licensee Event Report on Oconee
Philosophical Transactions of the Royal
Nuclear Station Unit Three, Report No.
Society London, B 298, pp. 199-209, 1982.
05000287/97-03-0. Washington, D.C., U.S.
Shannon, C. E., and Weaver, W., The Nuclear Regulatory Commission, 1997.
Mathematical Theory of Communication,
US NRC, Augmented Inspection Team Report on
Champaign, Illinois: University of Illinois
Davis Besse, Report No. 50-346/02-03.
Press, 1949.
Washington, D.C., U.S. Nuclear Regulatory
Shukri, T., and Mosleh, A., “A Dynamic PRA Commission, 2002.
Model: DS-IDA, Its Advantages and Possible
Veseley, W. E., et al. Fault Tree Handbook,
Applications,” Proceedings of the 4th
NUREG-0492. Washington, D.C., U.S.
International Conference on Probabilistic
Nuclear Regulatory Commission, 1981.
Safety and Management, New York,
Springer-Verlag Publishers, pp. 2667–2672, Wakefield, D., Parry, G.W., and Spurgin, A.J.,
1998. Revised Systematic Human Reliability
Analysis Procedure (SHARP1), EPRI TR-

77
 10711, Palo Alto, California, Electric Power California, New York, Institute of Electronic
Research Institute, 1990. and Electrical Engineers, 1988.
Weil, R., and Apostolakis, G.E., “A Methodology Williams, J., “Toward an Improved Evaluation
for the Prioritization of Operating Experience Analysis Tool for Users of HEART,”
in Nuclear Power Plants,” Reliability International Conference on Hazard
Engineering and Systems Safety, 74, pp. 23- Identification and Risk Analysis, Human
42, 2001. Factors and Human Reliability in Process
Safety, Orlando, Florida. 1992.
Wickens, C. D., and Hollands, J.G., Engineering
Psychology and Human Performance (3rd Yerkes, R. M., and Dodson, J. D., “The Relation
Ed.), Upper Saddle River, New Jersey, of Strength of Stimulus to Rapidity of Habit
Prentice Hall, 1999. Formation,” Journal of Comparative
Neurology and Psychology, 18, pp. 459–482,
Williams, J. C., “A Data-Based Method for
1908.
Assessing and Reducing Human Error to
Improve Operational Performance,” in
Proceedings of the IEEE 4th Conference on
Human Factors in Power Plants, Monterey,

78
 Appendix A
HRA Worksheets for At-Power

A-1
A-2
 HRA Worksheets for At-Power
SPAR HUMAN ERROR WORKSHEET

Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO (skip
Part I – Diagnosis; start with Part II – Action) Why?

PART I. EVALUATE EACH PSF FOR DIAGNOSIS

A. Evaluate PSFs for the Diagnosis Portion of the Task, If Any.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Barely adequate time (≈2/3 x nominal) 10
Nominal time 1
Extra time (between 1 and 2 x nominal and > 0.1
than 30 min)
Expansive time (> 2 x nominal and > 30 min) 0.01
Insufficient information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 2
Processes Nominal 1
Good 0.8
Insufficient Information 1
Rev 1 (1/20/04)
Reviewer: ___________

A-3
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x x x x x x x x =

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1

Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP =

Reviewer: ___________

A-4
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Time available is ≈ the time required 10
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Insufficient Information 1
Experience/ Low 3
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 5
Processes Nominal 1
Good 0.5
Insufficient Information 1

Reviewer: ___________
A-5
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x x x x x x x x =

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1

Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP =

Reviewer: ___________

A-6
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.

Pw/od = Diagnosis HEP _________ + Action HEP _________ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here:

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
10 a moderate least high.
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od

Calculate Pw/d using the appropriate values:

Pw/d = (1 + (_______ * _______))/ _______ =

Reviewer: ___________

A-7
A-8
 Appendix B
HRA Worksheets for LP/SD

B-1
B-2
 HRA Worksheets for LP/SD
SPAR Human Error Worksheet
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO
(skip Part I – Diagnosis; start with Part II – Action) Why?

PART I. EVALUATE EACH PSF FOR DIAGNOSIS

A. Evaluate PSFs for the Diagnosis Portion of the Task.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Barely adequate time (≈ 2/3 x nominal) 10
Nominal time 1
Extra time (between 1and 2 x nominal and > 0.1
30 min)
Expansive time > 2 x nominal & > 30 min 0.1 to 0.01
Insufficient Information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 2
Processes Nominal 1
Good 0.8
Insufficient Information 1
Rev 1 (1/20/04)

Reviewer: ___________
B-3
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x x x x x x x x =

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1

Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP =

Reviewer: ___________

B-4
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Time available is ≈ the time required 10
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Insufficient Information 1
Experience/ Low 3
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 5
Processes Nominal 1
Good 0.5
Insufficient Information 1

Reviewer: ___________

B-5
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________

Basic Event Context:

Basic Event Description:

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x x x x x x x x =

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1

Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP =

Reviewer: ___________
Plant: Initiating Event: Basic Event : ____________ Event Coder:___________
B-6
Basic Event Context:

Basic Event Description:

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.

Pw/od = Diagnosis HEP _________ + Action HEP _________ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here:

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
10 a moderate least high.
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od

Calculate Pw/d using the appropriate values:

Pw/d = (1 + (_______ * _______))/ _______ =

Reviewer: ___________

B-7
B-8
 Appendix C
Full Power Worksheets for SGTR Example

C-1
C-2
 Appendix C
Full Power Worksheets for SGTR Example
It is assumed that the reactor is at 100% power when the steam generator tube rupture (SGTR) occurs. Given an
SGTR, secondary cooling is required for decay heat removal provided a successful reactor trip has occurred. Early
core decay heat removal is required for a SGTR event. Successful operation of secondary cooling will start
depressurizing the RCS in order to isolate the ruptured steam generator. HPI is used to provide makeup flow to
replenish the lost RCS inventory. With HPI and secondary cooling operating, the RCS pressure needs to be reduced
below the steam generator relief valve pressure and the steam generator is isolated, then the plant is placed in a
stable condition using secondary cooling. If the ruptured steam generator cannot be isolated, then RCS pressure
must continue to be lowered in order for shutdown cooling (SDC) to be placed in operation for long-term cooling.
Plant stabilization given HPI failed can also be accomplished provided the RCS is depressurized and the steam
generator is rapidly isolated.

Feed and bleed cooling could be used to remove decay heat if secondary cooling (i.e., AFW and MFW) is
unavailable. For feed and bleed cooling, both PORVs are required to open and remove the decay heat and HPI is
required to provide the makeup flow. An operator is required to open the PORVs and PORV block valves if they
are closed. The operator controls the flow from the HPI pumps in order to slowly depressurize the RCS. Given the
successful operation of feed and bleed, long-term cooling using high-pressure recirculation (HPR) and containment
sump recirculation (CSR) is required. These success criteria are consistent within the PWR class G plants.

A number of human actions are associated with the event tree for a generic SGTR at this plant. These include
failing to: diagnose, depressurize RCS < SGRV, depressurize after SGRV lift, isolate SG, throttle HPI to reduce
pressure, initiate RCS depressurization, etc. Only three of these (failing to diagnose, failing to throttle HPI, and
failing to depressurize RCS) are presented here as generalized examples of how to apply SPAR-H worksheets.

C-3
C-4
 HRA Worksheets for At Power
SPAR HUMAN ERROR WORKSHEET

Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-DIAG Event Coder:_dk_______

Basic Event Context: Reactor at 100% Power

Basic Event Description: Operator Fails to Diagnose SGTR

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO (skip
Part I – Diagnosis; start with Part II – Action) Why? Operator must evaluate a set of parameters (e.g., pressure
differences between SGs, rates of increase or decrease of level, etc.). Indications can be masked.

PART I. EVALUATE EACH PSF FOR DIAGNOSIS

A. Evaluate PSFs for the Diagnosis Portion of the Task, If Any.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Based on plant operating characteristics, it is
Time Barely adequate time (≈2/3 x nominal) 10 assumed that the crew had adequate (nominal)
Nominal time 1 time.
Extra time (between 1 and 2 x nominal and > 0.1
than 30 min)
Expansive time (> 2 x nominal and > 30 min) 0.01
Insufficient information 1
Stress/ Extreme 5 It is assumed that stress will be higher than
Stressors High 2 normal.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Medium tube rupture is moderately complex
Moderately complex 2 (feed versus trying to maintain level control).
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10 Simulator training emphasizing diagnosis of
Training Nominal 1 SGTR is provided.
High 0.5
Insufficient Information 1
Procedures Not available 50 The EOPs are symptom based.
Incomplete 20
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 Plant-specific SGTR diagnosis simplified by
HMI Poor 10 having SG level indication and associated
Nominal 1 gauges available for comparison.
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 2 Based upon license examining or review, this
Processes Nominal 1 plant has good work processes.
Good 0.8
Insufficient Information 1
Rev 1 (1/20/04)
Reviewer: aw i
C-5
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-DIAG Event Coder:_dk_______

Basic Event Context: Reactor at 100% Power

Basic Event Description: Operator Fails to Diagnose SGTR

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x 1 x 2 x 2 x 0.5 x 0.5 x 1 x 1 x 0.8 = 0.008

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = 0.008

Reviewer: aw i

C-6
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-DIAG Event Coder:_dk_______

Basic Event Context: Reactor at 100% Power

Basic Event Description: Operator Fails to Diagnose SGTR

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Time available is ≈ the time required 10
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Insufficient Information 1
Experience/ Low 3
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 5
Processes Nominal 1
Good 0.5
Insufficient Information 1

Reviewer: aw i

C-7
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-DIAG Event Coder:_dk_______

Basic Event Context: Reactor at 100% Power

Basic Event Description: Operator Fails to Diagnose SGTR

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x x x x x x x x = N/A

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = N/A

Reviewer: aw i

C-8
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-DIAG Event Coder:_dk_______

Basic Event Context: Reactor at 100% Power

Basic Event Description: Operator Fails to Diagnose SGTR

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.008
Pw/od = Diagnosis HEP _0.008___ + Action HEP __0____ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here: First task in sequence

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
least high.
10 a moderate
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero U

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od U

Calculate Pw/d using the appropriate values:

N/A
Pw/d = (1 + (_______ * _______))/ _______ =

Reviewer: aw I
C-9
C-10
 HRA Worksheets for At Power
SPAR HUMAN ERROR WORKSHEET

Plant: Plant A Initiating Event: SGTR Basic Event : HPI-XHE-XM-THTL Event Coder:_dk____

Basic Event Context: HEP2- Reactor at 100% Power

Basic Event Description: Operator Fails to Throttle HPI to Reduce RCS Pressure
Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO (skip
Part I – Diagnosis; start with Part II – Action) Why? Task directed by procedure. Involves turning off pumps or
closing down on throttle valve.

PART I. EVALUATE EACH PSF FOR DIAGNOSIS

A. Evaluate PSFs for the Diagnosis Portion of the Task, If Any.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Barely adequate time (≈2/3 x nominal) 10
Nominal time 1
Extra time (between 1 and 2 x nominal and > 0.1
than 30 min)
Expansive time (> 2 x nominal and > 30 min) 0.01
Insufficient information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 2
Processes Nominal 1
Good 0.8
Insufficient Information 1
Rev 1 (1/20/04)
Reviewer: aw i
C-11
Plant: Plant A Initiating Event: SGTR Basic Event : HPI-XHE-XM-THTL Event Coder:_dk____

Basic Event Context: HEP2- Reactor at 100% Power

Basic Event Description: Operator Fails to Throttle HPI to Reduce RCS Pressure

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x x x x x x x x = N/A

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = N/A

Reviewer: aw i

C-12
Plant: Plant A Initiating Event: SGTR Basic Event : HPI-XHE-XM-THTL Event Coder:_dk____

Basic Event Context: HEP2- Reactor at 100% Power

Basic Event Description: Operator Fails to Throttle HPI to Reduce RCS Pressure

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 For a medium break, approximately 4-5
Time Time available is ≈ the time required 10 minutes or less is expected to be the time
Nominal time 1 available, but the action only takes a minute to
Time available ≥ 5x the time required 0.1 perform.
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5 You know what the problem is now, but the
Stressors High 2 situation remains stressful. Taking action
Nominal 1 reduces some of the stress you had under
Insufficient Information 1 diagnosis.
Complexity Highly complex 5 There is more than one pump in more than one
Moderately complex 2 train. May also have to bypass interlocks.
Nominal 1
Insufficient Information 1
Experience/ Low 3 You seem to get SI with almost every event and
Training Nominal 1 the crew must deal with it. You do it all the
High 0.5 time.
Insufficient Information 1
Procedures Not available 50 Expected and trained to do it from memory and
Incomplete 20 then check against procedure.
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 Mimics are good for this. Controls are well
HMI Poor 10 labeled. The presentation of the two trains is
Nominal 1 well laid out. PZR pressure and Rx level can
Good 0.5 be referenced by the crew.
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 5 Determined on the basis of analyst evaluation
Processes Nominal 1 of plant specific information.
Good 0.5
Insufficient Information 1

Reviewer: aw i
C-13
Plant: Plant A Initiating Event: SGTR Basic Event : HPI-XHE-XM-THTL Event Coder:_dk____

Basic Event Context: HEP2- Reactor at 100% Power

Basic Event Description: Operator Fails to Throttle HPI to Reduce RCS Pressure

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x 1 x 2 x 2 x 0.5 x 1 x 0.5 x 1 x 0.5 = 0.0005

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = 0.0005

Reviewer: aw i

C-14
Plant: Plant A Initiating Event: SGTR Basic Event : HPI-XHE-XM-THTL Event Coder:_dk____

Basic Event Context: HEP2- Reactor at 100% Power

Basic Event Description: Operator Fails to Throttle HPI to Reduce RCS Pressure

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.0005
Pw/od = Diagnosis HEP __0______ + Action HEP _0.0005___ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here: First HEP is failure to
diagnose an SGTR. It is not necessary to diagnose SGTR to reach the need to throttle HPI. You will be
directed to throttle HPI from some other emergency operating procedures (EOP) if not the SGTR EOP.

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
10 a moderate least high.
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero U

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od U

Calculate Pw/d using the appropriate values:

N/A
Pw/d = (1 + (_______ * _______))/ _______ =

Reviewer: aw i
C-15
C-16
 HRA Worksheets for At Power
SPAR HUMAN ERROR WORKSHEET

Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-XM-SG Event Coder:_dk_____

Basic Event Context: Preceded by Failure to Throttle HPI

Basic Event Description: Failure to Initiate RCS Depressurization

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO (skip
Part I – Diagnosis; start with Part II – Action) Why? This task involves careful control rather than diagnosis.
Elements of diagnosis may be present within individual operator actions as the procedure is followed, but the
procedure is prescriptive.
PART I. EVALUATE EACH PSF FOR DIAGNOSIS
A. Evaluate PSFs for the Diagnosis Portion of the Task, If Any.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Barely adequate time (≈2/3 x nominal) 10
Nominal time 1
Extra time (between 1 and 2 x nominal and > 0.1
than 30 min)
Expansive time (> 2 x nominal and > 30 min) 0.01
Insufficient information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 2
Processes Nominal 1
Good 0.8
Insufficient Information 1
Rev 1 (1/20/04)
Reviewer: aw i
C-17
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-XM-SG Event Coder:_dk_____

Basic Event Context: Preceded by Failure to Throttle HPI

Basic Event Description: Failure to Initiate RCS Depressurization

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x x x x x x x x = N/A

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = N/A

Reviewer: aw i

C-18
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-XM-SG Event Coder:_dk_____

Basic Event Context: Preceded by Failure to Throttle HPI

Basic Event Description: Failure to Initiate RCS Depressurization

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Any leak that will pop a relief valve is time
Time Time available is ≈ the time required 10 critical.
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5 You’ve breached a containment barrier and are
Stressors High 2 concerned about a second barrier (relief
Nominal 1 valves).
Insufficient Information 1
Complexity Highly complex 5 Not a 1-person evolution—1 in charge and 2
Moderately complex 2 workers is barely adequate—often failed on
Nominal 1 exam—always failed the first time a team of 3
attempts it.
Insufficient Information 1
Experience/ Low 3 Lots of training on this.
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50 Sufficient guidelines exist.
Incomplete 20
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 Task takes place all over control room—
HMI Poor 10 requires time sharing between tasks.
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 5 Plant specific for this facility as determined by
Processes Nominal 1 review by license examiners.
Good 0.5
Insufficient Information 1

Reviewer: aw i
C-19
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-XM-SG Event Coder:_dk_____

Basic Event Context: Preceded by Failure to Throttle HPI

Basic Event Description: Failure to Initiate RCS Depressurization

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x 10 x 5 x 5 x 0.5 x 1 x 1 x 1 x 0.5 = 0.0625

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
0.0589
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = 0.0589

Reviewer: aw i

C-20
Plant: Plant A Initiating Event: SGTR Basic Event : RCS-XHE-XM-SG Event Coder:_dk_____

Basic Event Context: Preceded by Failure to Throttle HPI

Basic Event Description: Failure to Initiate RCS Depressurization

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.0589
Pw/od = Diagnosis HEP ____0____ + Action HEP __0.0589__ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here:

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low U If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
least high.
10 a moderate
11 d na moderate
Although this is the third event and
12 a moderate
would normally have at least moderate
13 nc s na low dependency, the previous event had zero
14 a low dependency. This event is therefore
15 d na low treated as the second event following a
16 a low zero dependency prior event.
17 zero

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20 U
For Zero Dependence the probability of failure is Pw/od

Calculate Pw/d using the appropriate values:

0.105955
Pw/d = (1 + ( 19 * 0.0589 ))/ 20 =

Reviewer: aw i

C-21
C-22
 Appendix D
LP/SD Worksheets For PWR LOI with RCS Pressurized

D-1
D-2
 Appendix D
LP/SD Worksheets for PWR Loss of Inventory (LOI) with RCS
Pressurized
The scenario evaluated in this report makes use of a low power and shutdown (LP/SD) standardized plant analysis
risk model for a U.S. PWR nuclear plant. Specifically, the model was derived from NUREG/CR-6144 (1994) and
the at-power operation model for the corresponding plant. The model is organized around a number of plant
operating states likely to occur during either (a) refueling, (b) plant maintenance with drained reactor coolant
system, (c) nondrained maintenance that uses the RHR system for removal of decay heat, or (d) nondrained
maintenance without using the RHR system. Event trees, fault trees, and basic event data were compiled but are not
part of this report. The SPAR application in the following corresponds to HEPs that would be included as part of
SPAR basic events.
The scenario selected refers to a loss of inventory initiating event that leads to a reduction in RCS inventory that in
turn, leads to a loss of RHR. A loss of inventory event tree is presented as Table D-1. During the formal analysis,
the loss of inventory event tree was broken into two separate event trees because of differences in the initiating
events. One tree uses a demand related initiating event, the other an hourly-initiating event. The demand tree refers
to over draining events when the RCS is being reduced to mid-loop. The event tree reviewed for purposes of
SPAR-H refinement and application was from the hourly group, where loss of inventory occurs with the RCS
pressurized. One of the prominent events is the success or failure of RCS make-up by the operators. Success
implies that make-up water is being provided to the RCS by either one train of HPHPSI, both trains of CVCS, or
one train of the low-pressure injection (LPI). Success requires an operator to start and align the suction of the
injection pumps to the RWST and to align the discharge to the RCS cold legs. Similar considerations were made
when determining the HEPs for all basic events. Fault trees underlying individual basic events were also
determined but are not presented as part of this report. These trains include operator failures as well as component
unavailability information, and time window information usually represented at this level of analysis. As with any
event analysis, the HEP determined by the SPAR-H method only identifies the human error contribution to the
basic event frequency.

D-3
 LOSS OF OPERATOR RCS LOSS OF RHR SECONDARY FORCED BLEED LONG-TERM
INVENTORY DIAGNOSES INVENTORY INVENTORY RECOVERY COOLING FEED (SPILL) RECIRC
RCS EVENT MAKE-UP TERMINATED
PRESSURIZED

TW D M LOI-NT R S F B C # STATE NOTES

1 OK

2 OK

3 OK

4 X

5 X

F-HPI
6 X

7 OK

D-4
8 X

9 X

10 X

LOIG2 - TRANSFER - LOSS OF INVENTORY PRESSURIZED (POSG 2) 2004/01/21

Figure D-1. Loss of inventory event tree with RCS pressurized for a nuclear power plant.
 HRA Worksheets for LP/SD
SPAR Human Error Worksheet
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-DIAP2 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Operator Fails to Diagnose Loss of Inventory (1st Event)

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO
(skip Part I – Diagnosis; start with Part II – Action) Why?

PART I. EVALUATE EACH PSF FOR DIAGNOSIS

A. Evaluate PSFs for the Diagnosis Portion of the Task.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Given that an isolated leak rate is occurring.
Time Barely adequate time (≈ 2/3 x nominal) 10
Nominal time 1
Extra time (between 1and 2 x nominal and > 0.1
30 min)
Expansive time > 2 x nominal & > 30 min 0.1 to 0.01
Insufficient Information 1
Stress/ Extreme 5 Extreme stress is too dramatic for shutdown
Stressors High 2 activity. Some stress related to concern for
Nominal 1 cavitating the centrifical shutdown cooling
Insufficient Information 1 pumps.
Complexity Highly complex 5 If plant is at mid loop, operators are balancing
Moderately complex 2 water and air without a lot of inertia.
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10 Extensive training and experience.
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50 Emergency Operating Procedures (EOPs) are
Incomplete 20 symptom-based.
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 System less well designed for LP/SD activities.
HMI Poor 10 The reactor vessel level monitoring system is
Nominal 1 assumed to be available.
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 2 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.8 model and therefore nominal is an appropriate
Insufficient Information 1 choice.
Rev 1 (1/20/04)

Reviewer: aw i

D-5
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-DIAP2 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Operator Fails to Diagnose Loss of Inventory (1st Event)

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x 1 x 2 x 1 x 0.5 x 0.5 x 10 x 1 x 1 = 0.05

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = 0.05

Reviewer: aw i

D-6
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-DIAP2 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Operator Fails to Diagnose Loss of Inventory (1st Event)

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0
Time Time available is ≈ the time required 10
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5
Stressors High 2
Nominal 1
Insufficient Information 1
Complexity Highly complex 5
Moderately complex 2
Nominal 1
Insufficient Information 1
Experience/ Low 3
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50
Incomplete 20
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50
HMI Poor 10
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0
Duty Degraded Fitness 5
Nominal 1
Insufficient Information 1
Work Poor 5
Processes Nominal 1
Good 0.5
Insufficient Information 1

Reviewer: aw i

D-7
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-DIAP2 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Operator Fails to Diagnose Loss of Inventory (1st Event)

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x x x x x x x x = N/A

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = N/A

Reviewer: aw i

D-8
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-DIAP2 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Operator Fails to Diagnose Loss of Inventory (1st Event)

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.05
Pw/od = Diagnosis HEP _0.05_____ + Action HEP ___0_____ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here: This task is first in the event
tree; no previous human actions considered.

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
10 a moderate least high.
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero U

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od U

Calculate Pw/d using the appropriate values:

N/A
Pw/d = (1 + (_______ * _______))/ _______ =

Reviewer: aw i
D-9
D-10
 HRA Worksheets for LP/SD
SPAR Human Error Worksheet
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-LOI123 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Failure to Recover RHR

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO
(skip Part I – Diagnosis; start with Part II – Action) Why?

PART I. EVALUATE EACH PSF FOR DIAGNOSIS

A. Evaluate PSFs for the Diagnosis Portion of the Task.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Given than an isolatable leak rate is occurring.
Time Barely adequate time (≈ 2/3 x nominal) 10 Time to boil can vary as can time to uncover
Nominal time 1 the core.
Extra time (between 1and 2 x nominal and > 0.1
30 min)
Expansive time > 2 x nominal & > 30 min 0.1 to 0.01
Insufficient Information 1
Stress/ Extreme 5 Extreme stress is too dramatic for shutdown
Stressors High 2 activity.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Based on the generic model, it is assumed that
Moderately complex 2 the diagnosis task is not difficult and not
Nominal 1 especially complex.
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10 Extensive training and experience.
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50 EOPs are symptom based.
Incomplete 20
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 System less well-designed for LP/SD activities.
HMI Poor 10 Operators need to look at AMMeter oscillations
Nominal 1 as an indirect pump status indication.
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 2 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.8 model and therefore nominal is an appropriate
Insufficient Information 1 choice.
Rev 1 (1/20/04)
Reviewer: aw i

D-11
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-LOI123 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Failure to Recover RHR

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x 1 x 2 x 1 x 0.5 x 0.5 x 10 x 1 x 1 = 0.05

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = 0.05

Reviewer: aw i

D-12
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-LOI123 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Failure to Recover RHR

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 The time window afforded by plant operating
Time Time available is ≈ the time required 10 state may affect new response but by a factor of
Nominal time 1 no more than 1 to 2.
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5 Stress is assumed to decrease from the level
Stressors High 2 during diagnosis to be nominal when
Nominal 1 performing the action.
Insufficient Information 1
Complexity Highly complex 5 Based on the generic model, it is assumed that
Moderately complex 2 the action task is not difficult and not especially
Nominal 1 complex.
Insufficient Information 1
Experience/ Low 3 Crew is well trained on residual heat removal
Training Nominal 1 (RHR) for “at power” context.
High 0.5
Insufficient Information 1
Procedures Not available 50 Not as well developed for shutdown activities.
Incomplete 20
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 Once diagnosed, good support for corrective
HMI Poor 10 action; appropriate feedback is provided for this
Nominal 1 task.
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 5 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.5 model and therefore nominal is an appropriate
Insufficient Information 1 choice.

Reviewer: aw i

D-13
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-LOI123 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Failure to Recover RHR

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x 1 x 1 x 1 x 0.5 x 5 x 0.5 x 1 x 1 = 0.00125

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = 0.00125

Reviewer: aw i

D-14
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-LOI123 Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized

Basic Event Description: Failure to Recover RHR

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.05125
Pw/od = Diagnosis HEP _0.05____ + Action HEP __0.00125_ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here:

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate U sequence, then the dependency is at
least high.
10 a moderate
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7 U
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od

Calculate Pw/d using the appropriate values:

0.187
Pw/d = (1 + ( 6 * 0.05125 ))/ 7 =

Reviewer: aw i
D-15
D-16
 HRA Worksheets for LP/SD
SPAR Human Error Worksheet
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-XM-FB Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized/Failure to Start Pump

Basic Event Description: Failure to Force Feed

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO
(skip Part I – Diagnosis; start with Part II – Action) Why?

PART I. EVALUATE EACH PSF FOR DIAGNOSIS

A. Evaluate PSFs for the Diagnosis Portion of the Task.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Given that an isolatable leak rate is occurring.
Time Barely adequate time (≈ 2/3 x nominal) 10
Nominal time 1
Extra time (between 1and 2 x nominal and > 0.1
30 min)
Expansive time > 2 x nominal & > 30 min 0.1 to 0.01
Insufficient Information 1
Stress/ Extreme 5 Extreme stress is too dramatic for shutdown
Stressors High 2 activity.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Based on the generic model, it is assumed that
Moderately complex 2 the diagnosis task is not difficult and not
Nominal 1 especially complex.
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10 Extensive training and experience.
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50 EOPs are symptom based.
Incomplete 20
Available, but poor 5
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 System less well-designed for LP/SD activities.
HMI Poor 10 Key indication for the operators may include
Nominal 1 pressurizer level, reactor level, and general mid
Good 0.5 loop instrumentation.
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 2 Work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.8 model and therefore nominal is an appropriate
Insufficient Information 1 choice.
Rev 1 (1/20/04)
Reviewer: aw i

D-17
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-XM-FB Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized/Failure to Start Pump

Basic Event Description: Failure to Force Feed

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x 1 x 2 x 1 x 0.5 x 0.5 x 10 x 1 x 1 = 0.05

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = 0.05

Reviewer: aw i
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-XM-FB Event Coder:_dk_______
D-18
Basic Event Context: Loss of Inventory with RCS Pressurized/Failure to Start Pump

Basic Event Description: Failure to Force Feed

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Based upon the leak rate, ample time available
Time Time available is ≈ the time required 10 to resolve leak.
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5 Task action is easier than diagnosis; therefore,
Stressors High 2 nominal stress is assumed.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Moderately complex actions are required to
Moderately complex 2 force feed.
Nominal 1
Insufficient Information 1
Experience/ Low 3 Crew is well trained.
Training Nominal 1
High 0.5
Insufficient Information 1
Procedures Not available 50 Procedures are sufficient for the action
Incomplete 20 required.
Available, but poor 5
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 Once diagnosed, good support for corrective
HMI Poor 10 action; appropriate feedback provided for task.
Nominal 1
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 5 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.5 model and therefore nominal is an appropriate
Insufficient Information 1 choice.

Reviewer: aw i

D-19
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-XM-FB Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized/Failure to Start Pump

Basic Event Description: Failure to Force Feed

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x 0.1 x 1 x 2 x 0.5 x 1 x 1 x 1 x 1 = 0.0001

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = 0.0001

Reviewer: aw i

D-20
Plant: Plant B Initiating Event: LOI Basic Event : RHR-XHE-XM-FB Event Coder:_dk_______

Basic Event Context: Loss of Inventory with RCS Pressurized/Failure to Start Pump

Basic Event Description: Failure to Force Feed

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.0501
Pw/od = Diagnosis HEP _0.05_____ + Action HEP _0.0001__ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here:

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete U When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
least high.
10 a moderate
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1. U

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od

Calculate Pw/d using the appropriate values:

1.0
Pw/d = (1 + (_______ * _______))/ _______ =

Reviewer: aw i
D-21
D-22
 Appendix E
Worksheets for Dry Cask

E-1
E-2
 Appendix E
Worksheets for Dry Cask
The following three examples are SPAR-H applications for a screening HRA performed on dry cask storage
operations for spent commercial reactor fuel. The dry cask storage operation includes loading spent fuel assemblies
into a canister contained in a cement cask under water in the spent fuel pool, placing the lid with drain pipe
assembly on the canister, removing the cask from the pool, sealing the canister, drying and inserting the canister,
closing the cask, drying the cask annulus, and moving the cask to an outdoor storage pad.

The first example is the SPAR-H worksheet for the task of loading the fuel assemblies into the canister. The
potential error modeled is improper loading by placing a fuel assembly into a wrong location in the canister. A
loading map is provided to the crew. The map indicates specific spent fuel assemblies by serial number and the
specific placement location of each in the canister. The fuel crane operator selects, moves, and places each
assembly into the cask using a video image at his workstation on the crane from an underwater camera attached to
the cranes’ grapple assembly. Each fuel assembly’s serial number is stamped onto the top of the assembly.
Worksheet ratings that are other than nominal are “moderate complexity” and “poor” ergonomics for both the
diagnosis and action component of the task. Note that the worksheets do not account for latent errors related to the
production of the fuel-loading map.

The second example is the SPAR-H worksheet for the operators failing to properly remove water from the canister
by stepped vacuuming during the drying and inserting process. This activity takes place after the cask has been
removed from the fuel pool and the canister has been sealed. The diagnosis component and action component of the
activity each include a worksheet rating of “moderate complexity.” The diagnosis component complexity rating
considers the required calculation of the maximum time allowed for wet operations (e.g., allowable time from
removal from fuel pool to when the canister drying is completed). Both the diagnosis and the action component
complexity ratings reflect requirements for multiple valve line-ups, multiple timings for stepped drying, and
continuous monitoring and multiple recordings of canister pressure during drying to allow determination of drying
acceptance criteria.

The third example is the SPAR-H worksheet for operators failing to properly perform vacuum drying system
connections and set-up to enable drying of the cask annulus during the close cask phase of the operation. The
worksheet rating of complexity is “moderate complexity” for both the diagnosis and action components of the
activity. This reflects the multiple steps, components, connections, and manipulations required. The rating for
procedures is “available, but poor” for both the diagnosis and action components of the activity. This rating reflects
that the procedure refers to an attachment showing connections for the canister rather than the cask (which employs
different valve connections), and that the attachment has inconsistent or missing symbols.

E-3
E-4
 HRA Worksheets for LP/SD
SPAR Human Error Worksheet
Plant: Plant X Initiating Event: Fuel Overload Basic Event : XHE_____ Event Coder:_bf________

Basic Event Context: Fuel Overloading Resulting in Higher Surface Radiation than Allowable

Basic Event Description: Failure to Load Fuel Assemblies Properly

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO
(skip Part I – Diagnosis; start with Part II – Action) Why? Multiple assemblies/serial numbers/specific placement
locations/verifications.
PART I. EVALUATE EACH PSF FOR DIAGNOSIS
A. Evaluate PSFs for the Diagnosis Portion of the Task.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Adequate (nominal) time is assumed in this
Time Barely adequate time (≈ 2/3 x nominal) 10 example.
Nominal time 1
Extra time (between 1and 2 x nominal and > 0.1
30 min)
Expansive time > 2 x nominal & > 30 min 0.1 to 0.01
Insufficient Information 1
Stress/ Extreme 5 There is no information available on stress in
Stressors High 2 this example.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Multiple fuel assemblies, serial numbers
Moderately complex 2 stamped on the top of each assembly, specific
Nominal 1 placement locations in the canister.
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10 Average (nominal) experience and training are
Training Nominal 1 assumed in this example.
High 0.5
Insufficient Information 1
Procedures Not available 50 Procedures are plant specific. This analysis
Incomplete 20 was determined from a generic model and
Available, but poor 5 therefore nominal is an appropriate choice.
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 Performed by remote control underwater using
HMI Poor 10 video camera view. (Assembly selection,
Nominal 1 placement, and verification.)
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 2 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.8 model and therefore nominal is an appropriate
Insufficient Information 1 choice.
Rev 1 (1/20/04)
Reviewer: aw i
E-5
Plant: Plant X Initiating Event: Fuel Overload Basic Event : XHE_____ Event Coder:_bf________

Basic Event Context: Fuel Overloading Resulting in Higher Surface Radiation than Allowable

Basic Event Description: Failure to Load Fuel Assemblies Properly

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x 1 x 1 x 2 x 1 x 1 x 10 x 1 x 1 = 0.2

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = 0.2

Reviewer: aw i

E-6
Plant: Plant X Initiating Event: Fuel Overload Basic Event : XHE_____ Event Coder:_bf________

Basic Event Context: Fuel Overloading Resulting in Higher Surface Radiation than Allowable

Basic Event Description: Failure to Load Fuel Assemblies Properly

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Adequate (nominal) time is assumed in this
Time Time available is ≈ the time required 10 example.
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5 There is no information available on stress in
Stressors High 2 this example.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Multiple assemblies with specific placements in
Moderately complex 2 canister.
Nominal 1
Insufficient Information 1
Experience/ Low 3 Average (nominal) experience and training are
Training Nominal 1 assumed in this example.
High 0.5
Insufficient Information 1
Procedures Not available 50 Procedures are plant specific. This analysis
Incomplete 20 was determined from a generic model and
Available, but poor 5 therefore nominal is an appropriate choice.
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 Performed by remote control underwater using
HMI Poor 10 video camera view. (Assembly selections,
Nominal 1 placement, and verification serial number
Good 0.5 stamps with underwater camera.)
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 5 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.5 model and therefore nominal is an appropriate
Insufficient Information 1 choice.

Reviewer: aw i

E-7
Plant: Plant X Initiating Event: Fuel Overload Basic Event : XHE_____ Event Coder:_bf________

Basic Event Context: Fuel Overloading Resulting in Higher Surface Radiation than Allowable

Basic Event Description: Failure to Load Fuel Assemblies Properly

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x 1 x 1 x 2 x 1 x 1 x 10 x 1 x 1 = 0.02

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = 0.02

Reviewer: aw i

E-8
Plant: Plant X Initiating Event: Fuel Overload Basic Event : XHE_____ Event Coder:_bf________

Basic Event Context: Fuel Overloading Resulting in Higher Surface Radiation than Allowable

Basic Event Description: Failure to Load Fuel Assemblies Properly

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.22
Pw/od = Diagnosis HEP __0.2____ + Action HEP ___0.22___ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here: First event in sequence.

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
least high.
10 a moderate
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low
17 zero U

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od U

Calculate Pw/d using the appropriate values:

N/A
Pw/d = (1 + (_______ * _______))/ _______ =

Reviewer: aw i
E-9
E-10
 HRA Worksheets for LP/SD
SPAR Human Error Worksheet
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Dry and Insert the Canister

Basic Event Description: Operators Fail to Remove Water Properly by Stepped Vacuuming

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO
(skip Part I – Diagnosis; start with Part II – Action) Why? Maximum time calculation, timings, monitorings,
recordings.
PART I. EVALUATE EACH PSF FOR DIAGNOSIS
A. Evaluate PSFs for the Diagnosis Portion of the Task.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Adequate (nominal) time is assumed in this
Time Barely adequate time (≈ 2/3 x nominal) 10 example.
Nominal time 1
Extra time (between 1and 2 x nominal and > 0.1
30 min)
Expansive time > 2 x nominal & > 30 min 0.1 to 0.01
Insufficient Information 1
Stress/ Extreme 5 There is no information available on stress in
Stressors High 2 this example.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Maximum time calculations, multiple valve
Moderately complex 2 lineups, timings, and monitoring.
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10 Average (nominal) experience and training are
Training Nominal 1 assumed in this example.
High 0.5
Insufficient Information 1
Procedures Not available 50 Procedures are plant specific. This analysis
Incomplete 20 was determined from a generic model and
Available, but poor 5 therefore nominal is an appropriate choice.
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 The ergonomics are improved for this task
HMI Poor 10 compared to previous task, such that there are
Nominal 1 no indicators that ergonomics would
Good 0.5 significantly contribute to human error.
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 2 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.8 model and therefore nominal is an appropriate
Insufficient Information 1 choice.
Rev 1 (1/20/04)
Reviewer: aw i

E-11
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Dry and Insert the Canister

Basic Event Description: Operators Fail to Remove Water Properly by Stepped Vacuuming

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x 1 x 1 x 2 x 1 x 1 x 1 x 1 x 1 = 0.02

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = 0.02

Reviewer: aw i

E-12
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Dry and Insert the Canister

Basic Event Description: Operators Fail to Remove Water Properly by Stepped Vacuuming

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Adequate (nominal) time is assumed in this
Time Time available is ≈ the time required 10 example.
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5 There is no information available on stress in
Stressors High 2 this example.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Maximum time calculations, multiple valve
Moderately complex 2 lineups, timings, and monitoring.
Nominal 1
Insufficient Information 1
Experience/ Low 3 Average (nominal) experience and training are
Training Nominal 1 assumed in this example.
High 0.5
Insufficient Information 1
Procedures Not available 50 Procedures are plant specific. This analysis
Incomplete 20 was determined from a generic model and
Available, but poor 5 therefore nominal is an appropriate choice.
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 The ergonomics are improved for this task
HMI Poor 10 compared to previous task, such that there are
Nominal 1 no indicators that ergonomics would
Good 0.5 significantly contribute to human error.
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 5 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.5 model and therefore nominal is an appropriate
Insufficient Information 1 choice.

Reviewer: aw i

E-13
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Dry and Insert the Canister

Basic Event Description: Operators Fail to Remove Water Properly by Stepped Vacuuming

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x 1 x 1 x 2 x 1 x 1 x 1 x 1 x 1 = 0.002

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = 0.002

Reviewer: aw i

E-14
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Dry and Insert the Canister

Basic Event Description: Operators Fail to Remove Water Properly by Stepped Vacuuming

PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.022
Pw/od = Diagnosis HEP __0.02___ + Action HEP __0.002___ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here: Not the same crew, time,
location, and cues as previous event.

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
least high.
10 a moderate
11 d na moderate
12 a moderate
13 nc s na low
14 a low
15 d na low
16 a low U
17 zero

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7
For Low Dependence the probability of failure is (1+19 x Pw/od)/20 U
For Zero Dependence the probability of failure is Pw/od

Calculate Pw/d using the appropriate values:

0.0709
Pw/d = (1 + ( 19 * 0.022 ))/ 20 =

Reviewer: aw i
E-15
E-16
 HRA Worksheets for LP/SD
SPAR Human Error Worksheet
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Cask Closure

Basic Event Description: Operators Fail to Perform Vacuum Drying System Connections/Setup

Does this task contain a significant amount of diagnosis activity? YES (start with Part I–Diagnosis) NO
(skip Part I – Diagnosis; start with Part II – Action) Why? Multiple steps, components, connections, and
manipulations.
PART I. EVALUATE EACH PSF FOR DIAGNOSIS
A. Evaluate PSFs for the Diagnosis Portion of the Task.
PSFs PSF Levels Multiplier for Please note specific reasons for
Diagnosis PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Adequate (nominal) time is assumed in this
Time Barely adequate time (≈ 2/3 x nominal) 10 example.
Nominal time 1
Extra time (between 1and 2 x nominal and > 0.1
30 min)
Expansive time > 2 x nominal & > 30 min 0.1 to 0.01
Insufficient Information 1
Stress/ Extreme 5 There is no information available on stress in
Stressors High 2 this example.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Multiple steps, components, connections, and
Moderately complex 2 manipulations.
Nominal 1
Obvious diagnosis 0.1
Insufficient Information 1
Experience/ Low 10 Average (nominal) experience and training are
Training Nominal 1 assumed in this example.
High 0.5
Insufficient Information 1
Procedures Not available 50 Attachment refers to canister rather than cask,
Incomplete 20 which employs different valve connections and
Available, but poor 5 contains inconsistent or missing symbols.
Nominal 1
Diagnostic/symptom oriented 0.5
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 There are no indications that ergonomics would
HMI Poor 10 significantly contribute to human error for this
Nominal 1 task.
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 2 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.8 model and therefore nominal is an appropriate
Insufficient Information 1 choice.
Rev 1 (1/20/04)
Reviewer: aw i

E-17
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Cask Closure

Basic Event Description: Operators Fail to Perform Vacuum Drying System Connections/Setup

B. Calculate the Diagnosis Failure Probability.

(1) If all PSF ratings are nominal, then the Diagnosis Failure Probability = 1.0E-2
(2) Otherwise, the Diagnosis Failure Probability is: 1.0E-2 x Time x Stress or Stressors x Complexity x Experience
or Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Diagnosis: 1.0E-2x 1 x 1 x 2 x 1 x 5 x 1 x 1 x 1 = 0.1

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-2 for Diagnosis. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Diagnosis HEP with Adjustment Factor =

D. Record Final Diagnosis HEP.

If no adjustment factor was applied, record the value from Part B as your final diagnosis HEP. If an adjustment factor was applied, record
the value from Part C.

Final Diagnosis HEP = 0.1

Reviewer: aw i

E-18
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Cask Closure

Basic Event Description: Operators Fail to Perform Vacuum Drying System Connections/Setup

Part II. EVALUATE EACH PSF FOR ACTION

A. Evaluate PSFs for the Action Portion of the Task, If Any.

PSFs PSF Levels Multiplier for Please note specific reasons for
Action PSF level selection in this
column.
Available Inadequate time P(failure) = 1.0 Adequate (nominal) time is assumed in this
Time Time available is ≈ the time required 10 example.
Nominal time 1
Time available ≥ 5x the time required 0.1
Time available is ≥ 50x the time required 0.01
Insufficient Information 1
Stress/ Extreme 5 There is no information available on stress in
Stressors High 2 this example.
Nominal 1
Insufficient Information 1
Complexity Highly complex 5 Observers of this task suggest moderate
Moderately complex 2 complexity for operators.
Nominal 1
Insufficient Information 1
Experience/ Low 3 Average (nominal) experience and training are
Training Nominal 1 assumed in this example.
High 0.5
Insufficient Information 1
Procedures Not available 50 Attachment refers to canister rather than cask,
Incomplete 20 which employs different valve connections and
Available, but poor 5 contains inconsistent or missing symbols.
Nominal 1
Insufficient Information 1
Ergonomics/ Missing/Misleading 50 There are no indications that ergonomics would
HMI Poor 10 significantly contribute to human error for this
Nominal 1 task.
Good 0.5
Insufficient Information 1
Fitness for Unfit P(failure) = 1.0 The fitness for duty of the crew is plant and
Duty Degraded Fitness 5 crew specific. This analysis was determined
Nominal 1 from a generic model and therefore nominal is
Insufficient Information 1 an appropriate choice.
Work Poor 5 The work processes are plant and crew specific.
Processes Nominal 1 This analysis was determined from a generic
Good 0.5 model and therefore nominal is an appropriate
Insufficient Information 1 choice.

Reviewer: aw i

E-19
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Cask Closure

Basic Event Description: Operators Fail to Perform Vacuum Drying System Connections/Setup

B. Calculate the Action Failure Probability.

(1) If all PSF ratings are nominal, then the Action Failure Probability = 1.0E-3
(2) Otherwise, the Action Failure Probability is: 1.0E-3 x Time x Stress or Stressors x Complexity x Experience or
Training x Procedures x Ergonomics or HMI x Fitness for Duty x Processes
Action: 1.0E-3x 1 x 1 x 2 x 1 x 1 x 5 x 1 x 1 = 0.01

C. Calculate the Adjustment Factor IF Negative Multiple (≥3) PSFs are Present.

When 3 or more negative PSF influences are present, in lieu of the equation above, you must compute a composite
PSF score used in conjunction with the adjustment factor. Negative PSFs are present anytime a multiplier greater
than 1 is selected. The Nominal HEP (NHEP) is 1.0E-3 for Action. The composite PSF score is computed by
multiplying all the assigned PSF values. Then the adjustment factor below is applied to compute the HEP:

NHEP ⋅ PSFcomposite
HEP =
NHEP ⋅ (PSFcomposite − 1) + 1
N/A
Action HEP with Adjustment Factor =

D. Record Final Action HEP.

If no adjustment factor was applied, record the value from Part B as your final action HEP. If an adjustment factor was applied, record the
value from Part C.

Final Action HEP = 0.01

Reviewer: aw i

E-20
Plant: Plant X Initiating Event: Excess Water Criticality Basic Event : XHE__ Event Coder:_bf__

Basic Event Context: Cask Closure

Basic Event Description: Operators Fail to Perform Vacuum Drying System Connections/Setup
PART III. CALCULATE TASK FAILURE PROBABILITY WITHOUT FORMAL DEPENDENCE (PW/OD)

Calculate the Task Failure Probability Without Formal Dependence (Pw/od) by adding the Diagnosis Failure
Probability from Part I and the Action Failure Probability from Part II. In instances where an action is required
without a diagnosis and there is no dependency, then this step is omitted.
0.11
Pw/od = Diagnosis HEP __0.1____ + Action HEP _0.01____ =

Part IV. DEPENDENCY

For all tasks, except the first task in the sequence, use the table and formulae below to calculate the Task Failure
Probability With Formal Dependence (Pw/d).

If there is a reason why failure on previous tasks should not be considered, such as it is impossible to take the
current action unless the previous action has been properly performed, explain here: Not the same crew, time,
location, and cues as previous event.

Dependency Condition Table

Condition Crew Time Location Cues Dependency Number of Human Action Failures Rule
Number (same or (close in time (same or (additional or - Not Applicable.
different) or not close different) no Why?_________________
in time) additional)
1 s c s na complete When considering recovery in a series
2 a complete e.g., 2nd, 3rd, or 4th checker
3 d na high
4 a high If this error is the 3rd error in the
5 nc s na high sequence, then the dependency is at
6 a moderate least moderate.
7 d na moderate
8 a low If this error is the 4th error in the
9 d c s na moderate sequence, then the dependency is at
least high.
10 a moderate
11 d na moderate
Note that this is the 3 rd error in the
12 a moderate
sequence and defaults to moderate
13 nc s na low dependency.
14 a low
15 d na low
16 a low U
17 zero

Using Pw/od = Probability of Task Failure Without Formal Dependence (calculated in Part III):

For Complete Dependence the probability of failure is 1.

For High Dependence the probability of failure is (1+ Pw/od)/2
For Moderate Dependence the probability of failure is (1+6 x Pw/od)/7 U
For Low Dependence the probability of failure is (1+19 x Pw/od)/20
For Zero Dependence the probability of failure is Pw/od

Calculate Pw/d using the appropriate values:

0.2371
Pw/d = (1 + ( 6 * 0.11 ))/ 7 =

Reviewer: aw i
E-21
E-22
 Appendix F
Operational Examples of SPAR-H Method Assignment
of PSF Levels

F-1
F-2
 Appendix F
Operational Examples of SPAR-H Method Assignment of PSF
Levels
Available Time – Time available ≈ time required

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
HNP1 06/27/93 AIT 3 events occurred (06/22-06/27) while conducting tests. 6/22 & Shift supervisor felt he was
6/26: total loss of offsite power from wiring error & blown fuse under time pressure to process
respectively. 6/27: temporary loss of motor-control-center-5, the notification within the 12
which provides power to ECCS. Erroneous alert changed to minutes required by the
unusual. procedure (takes 10 minutes to
input the data) and did so at
the expense of assuring
information accuracy.

Stress/Stressors – High

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
ZIS1 02/21/97 AIT Following 48 hr limiting condition of operation (LCO) and 4 hr 39 people in control room
Technical Specification (TS) shutdown statement, Nuclear Station envelope with 15 people in
Operator (NSO) initiates improper control rod manipulations immediate vicinity of the
during unit 1 shutdown, inserts rods for 3'43", then rods withdrawn primary NSO operating rods
for 1'45" without estimated critical position calculation while and the US, high ambient
reactor (RX) substantially subcritical. RX tripped for containment noise level, attempts to restore
sump (CS) pump problem prior to criticality. Inadequate reactivity the 1CS pump was the most
management. intrusive activity during the
event.
NMP2 08/13/91 IIT Internal failure in main transformer caused turbine trip and RX Stress and time pressure were
scram. Degraded voltage resulted in simultaneous common-mode high. Event occurred just
loss of 5 uninterruptible power supplies to important control room before shift change. Operators
instrumentation and other plant equipment. Brought to safe had confidence in their
shutdown training.
NAS2 04/16/93 HPS Control problem in main generator voltage regulator led to Stress due to unfamiliar crew
overexcited condition and reactor trip. Auxiliary feedwater(AFW) composition. Sense of less
pumps disabled for eighteen minutes during reactor trip recovery. communication/feedback than
usual. Operator broke glass
cover on control board
indicator. Feeling of urgency.
EFP2 08/13/93 HPS Spurious reactor scram, loss of gland seal steam and condenser Numerous failures and the
vacuum resulted in main steam isolation valve (MSIV) closure and smell of smoke during the
steam relief valve (SRV) pressure control. initial stages of recovery
diverted or consumed operator
attention. Stress from
unexpected alarms, trips,
uncertainty of cause, and the
first RX trip at high power for
the crew.

F-3
Complexity – Moderately complex

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
NEE3 07/03/91 ASPLER Operator left demineralizer bypass valves shut in manual. Later, Control room operator (CRO)
instrument air to condensate demineralizer valve controller was performing multiple,
became blocked. All automatic valves shut, bypass could not concurrent tasks. Operator was
respond. Condensate booster pumps tripped. Main feed pumps interrupted by phone call
tripped. Reactor tripped. related to second task and
forgot to place the bypass
valve control in AUTO status
after interruption.

Experience/Training – Low

Event ID Event Summary PSF Description

(Plant/Date /
Report Type)
DCC1 09/12/95 In Mode 6 and defueled, West Centrifugal Charging Pump (CPP) Root cause was lack of
ASPLER tripped after 7 min. of a surveillance test. Pump had tripped due to requalification training, which
incorrect setting for overcurrent relay. Later determined pump had resulted in the calibration error
been inoperable for 6 mo., since last calibration. Personnel had made by the technicians.
used wrong technique resulting in miscalibration. Although adequately trained
initially, a significant amount
of time had elapsed since
training and a lack of
requalification training led to
the personnel error.
MGS1 06/26/90 Diesel generator 1A failed to reach required voltage in required Root cause of inappropriate
ASPLER time during operability test. Subsequent start attempt resulted in action by maintenance
valid failure due to unsuccessful loading attempt. Paint overspray personnel of painting area
was found on the commutator ring and fuel racks of diesel above fuel racks was
generator (DG) 1A and 1B. Both DGs were declared inoperable. unauthorized; maintenance
Paint removed from D/Gs and operability tests were successful. personnel relied on their own
experience as to what to paint.
Also Operations support
person in charge of D/Gs
believed (wrongly) specific
guidance about what to paint
was not necessary because the
same personnel had previously
painted the Unit 2 D/Gs.
BRF2 05/11/93 HPS Isolation of valve associated with indicator used to monitor & No training: crew had little
control pressure resulted in actions causing high pressure in experience with the tests
reactor coolant system (RCS) & an ARI/RPT engineered safety because they were performed
features (ESF) actuation during test infrequently. No simulator
training on test.
WGS3 06/10/95 AIT RX trip resulted from offsite electrical disturbance (lightning Inadequate training of
arrester failure). Fire in turbine building switchgear room resulted operators to respond to initial
from auto load transfer problems. Shutdown cooling delayed by indications of potentially
failure of isolation valves for both shutdown cooling trains. significant fire. Fire brigade
training weakness resulted in
reluctance to use water to
extinguish fire when other fire
suppression methods failed.

F-4
 Event ID Event Summary PSF Description
(Plant/Date /
Report Type)
PAV3 05/04/92 AIT Loss of non-safety related annunciator and computer alarm Not provided for loss of all
systems following a circuit breaker trip alarm verification that non-safety related
created an inadvertent short circuit. annunciators during normal or
abnormal operating conditions
because of perceived low
probability of such an event.
EFP2 08/13/93 HPS Spurious reactor scram, loss of gland seal steam and condenser No training: simulator training
vacuum resulted in MSIV closure and SRV pressure control. was not updated to reflect
manual control of the gland
seal steam system. No training
on how extra RO should assist
during event. No multiple
operator training.

Experience/Training – Nominal

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
WGS3 06/24/91 HPS After RX trip & power cutback, operator stabilized plant and while Timely response of control
reducing power, startup feedwater regulating valve failed open and room operators due to
caused increased level in steam generator (SG)2. Operator knowledge and training in
scrammed RX & initiated MSIV trip to prevent excess cooldown procedures and operating
from failed open safety block control valve (SBCV). principles. Crew had just
completed ten days refresher
training.
WCS1 09/23/91 AIT Loss of spent fuel pool level and cooling, loss of gate boot seals. Operators' training and
Breaker trip and associated loss of bus pao1.RCS transient induced familiarity with the plant were
by loss of 2 of 4 operating reactor coolant pumps (RCP)'s during assets in coping with the
solid plant operations gave rapid decrease in RCS pressure & RHR event.
heat sink.

Experience/Training – High

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
FCS1 07/03/92 HPS Loss of non-safety-related electrical converter led to a high Plant specific simulator
pressure RX trip followed by a partially failed open safety relief training helped in ability to
valve. Similar event had occurred here 6 years before. respond. Trained on LOCA's
and loss of inverter scenarios.
Also trained on implementing
emergency plan. Could be
improved to include actions
for degradation or failure of
computer system.
FCS1 07/03/92 AIT Loss of load event occurred resulting in RX trip & loss of coolant Simulator training received
event. Turbine control valves shut and pressure increased resulting was a significant factor in
in pressurizer code safety valve and uncontrolled loss of coolant. event mitigation. Loss of
coolant events included in
simulator training. Site
specific simulator has
provided increased training
time & procedure confidence.
Emergency planning practiced
weekly.

F-5
 Event ID Event Summary PSF Description
(Plant / Date /
Report Type)
PAV3 02/04/93 HPS A main feedwater pump high vibration annunciator alarmed while Combined crew experience
operating at 100% power. Safety injection initiated. RX and training were above the
automatically tripped on low steam generator levels one minute industry norm and contributed
later. to successful performance,
however there was no training
on conditions of this event.
Previous training included
command and control.
Simulator training was useful.

Procedures – Not Available

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
CAY1 10/17/92 AIT Loss of main control room annunciators following power supply Procedures not available for
loss. loss of annunciators. ‘Loss of
Plant Computer” procedures
not used. No list of which
alarms were on which power
supplies.
MNS3 12/31/90 AIT Catastrophic failure of two 6-in diameter pipes associated with the No administrative procedure
plant moisture separator drain system allowed a significant amount for evaluating through-wall
of hot condensate system water and steam to be released into the leaks in the failed system.
turbine building. Plant process computer lost.

Procedures – Available, but poor

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
NEE3 05/03/97 AIT Degradation of the high pressure injection system during unit Shutdown/cooldown
cooldown. Potential damage occurred to 2 of 3 high pressure procedure didn't guide
injection pumps. Letdown storage tank level and related suction sensitivity to RCS & systems
head failed to be maintained. Let down storage tank (LDST) inventory balancing during
erroneously indicated normal level, while actual inventory cooldown. Procedures
decreasing. provided limited assistance
because of non-awareness that
letdown storage tank
indications were inaccurate.
Operations Management
Procedure sent mixed
messages that perhaps
procedures were weak and
compliance not required.
IPS3 10/04/90 AIT Two fuel assemblies were inadvertently lifted out of the core with Procedure for fuel movement
the reactor upper internals during preparations for defueling. AIT deficient. Did not contain
concluded that guide pins were bent during may 1989 refueling. detailed information needed
on video inspection of
assemblies and positioning.
Problems with complicated
measuring requirements.
Format used notes
inappropriately; directions in
notes & note in wrong place.

F-6
Procedures – Nominal

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
LSC2 08/27/92 AIT Main turbine trip and subsequent scram due to thrust bearing Good use of procedures
failure indication numerous equipment problems followed scram. assisted in prioritization and
addressing individual
equipment problems.
PNP1 03/26/93 AIT Non-safety related 30-inch service water pipe break and Good use of procedures
subsequent flooding in some plant areas required a rapid reactor assisted in prioritization and
shutdown, including a manual scram, and consequent activation of combating of individual
safety equipment. Cause of small leak, enlarged by erosion, equipment problems.
unknown.

Ergonomics/HMI – Missing/Misleading

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
SGS2 12/13/92 AIT Loss of control room overhead annunciator system for 1-1/2 hours Operators not aware that
without knowledge of or response by operating staff. computer locked up and
annunciators were not
working. Failure mode not
readily detectable and
alternate alert not provided.
No human factors review of
remote configuration
workstation, which lacked
human factors features.
PIN2 02/20/92 AIT Residual heat removal system interruption due to over draining of Design of the electronic level
RX coolant system while attempting to establish stable mid-loop measurement instruments was
operation conditions, shutting off inservice RHR pump and incompatible with the nitrogen
interrupting heat removal. pressures specified in the
draindown procedure.
Instruments were essentially
unavailable.

Ergonomics/HMI – Poor

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
FCS1 07/03/92 HPS Loss of nonsafety-related electrical converter led to a high Arrangement/placement:
pressure RX trip followed by a partially failed open safety relief related displays & controls
valve. Similar event had occurred here 6 years before. were located at some distance
from each other. Difficulty in
obtaining info for failed
computer displays. HPSI valve
did not have consistent linear
controls.
NEE3 03/08/91 LER Rev 0. While shutdown during refueling, spilled 14,000 g of water Incorrect handwritten label in
from RCS & borated water storage to RX building during valve RX building emergency sump
test. Blank flange installed on wrong suction train. Not on isolation identifying wrong low
valve tested. Interrupted decay heat removal for 18.5 min. pressure injection suction
pipe. No formal labeling.

Ergonomics/HMI – Nominal

F-7
 Event ID Event Summary PSF Description
(Plant / Date /
Report Type)
PAV3 02/04/93 LER Rev 0. RX trip due to SG2 water level reaching low RPS trip set No unusual characteristics of
point following loss of main feedwater pump A, followed by the work location (e.g., noise,
multiple ESF actuations. Event diagnosed as an uncomplicated RX heat, poor lighting) directly
trip. contributed to this event.

Fitness for Duty – Degraded Fitness

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
HBR2 11/14/93 AIT Mismatch between actual power and power range nuclear Long vendor shifts and
instrumentation during startup, due to fuel assembly error by personnel illness contributed
vendor and operators lack of understanding of core geometry. to breaking or failing to notice
Power increase caused violation of tech specs, flux tilt & power damage to a fuel inspection
level anomalies. tool, resulting in loose parts in
control rod guide tube.

Work Processes – Poor

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
PAV3 05/04/92 AIT Loss of non-safety related annunciator and computer alarm Problems with work order
systems following a circuit breaker trip alarm verification that control & personnel safety
created an inadvertent short circuit. (electricians failed to use
safety equipment, adequate
safety precautions not taken).
Program barriers not in place
for engineering review of
modified work orders.
Inconsistent with management
expectations.
ZIS1 02/21/97 AIT Following 48 hr LCO and 4 hr TS shutdown statement, Nuclear Breakdown in command and
Station Operator (NSO) initiates improper control rod control, failure of ops
manipulations during Unit 1 shutdown, inserts rods for 3'43", then supervision to properly
rods withdrawn for 1'45" without estimated critical position exercise oversight
calculation while RX substantially subcritical. RX tripped for core responsibilities for ensuring
spray (CS) pump problem prior to criticality. Inadequate reactivity shift activities conducted in
management. controlled manner. Shutdown
(SD) briefing was informal,
poorly planned, ineffective.
"Event was primarily the
result of breakdown in
command and control."

F-8
 Event ID Event Summary PSF Description
(Plant / Date /
Report Type)
WNP2 04/09/95 AIT Reactor water cleanup valve was operated in violation of Inadequate communications
procedure cautions and requirements (prohibiting opening of the between control room
valve above 125 psig) while attempting to control reactor water supervisor (CRS) and shift
level during hot shutdown. manager. CRS didn't pay
attention to operator concerns,
communication was informal
and directions were vague.
Relief CRO was not informed
of valve position. Valve
position not recorded in
control room log.
Inadequate organizational
culture. Poor personal work
standards were root causes of
the event. Management
response to prior interpersonal
problems of the effected crew
was slow.

Work Processes – Nominal

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
PBS3 01/28/90 HPS Loss of electro hydraulic control (ECH) and resultant rapid Competent and constructive
shutdown due to o-ring failure in main turbine control valve. communications. Furthered
crew ability to function
effectively under trying
circumstances.
PAV3 05/04/92 AIT Loss of non-safety related annunciator and computer alarm Lack of intrusive supervisory
systems following a circuit breaker trip alarm verification that involvement in the initiation
created an inadvertent short circuit. and performance of routine
balance-of-plant electrical
work. Successful coordination
of short-term corrective
actions. Effectively avoided
challenges to plant safety
systems.

Work Processes – Good

Event ID Event Summary PSF Description

(Plant / Date /
Report Type)
PAV3 02/04/93 HPS A main feed water pump high vibration annunciator alarmed while Good command & control.
operating at 100% power. Safety injection initiated. RX Shift supervisor (SS) moved
automatically tripped on low steam generator levels one minute people out of control room,
later. had good overview, and was
out of operator’s way, yet
readily available to crew.
Emergency coordinator duties
transferred to available
qualified person, enhancing
SS oversight ability.

F-9
 Event ID Event Summary PSF Description
(Plant / Date /
Report Type)
NMP2 03/23/92 AIT 1 of 2 lines supplying off site power to Unit 2 inadvertently SS exhibited good command
tripped, causing loss of control room annunciators. Second trip and control while conducting
(power line) led to total loss of offsite power. One of two running plant restoration.
emergency diesel generators tripped due to loss of cool water.

F-10
 Appendix G
The Relative Relationship Among SPAR-H PSFs

G-1
G-2
 Appendix G
The Relative Relationship Among SPAR-H PSFs
Table G-1. The relative relationship among SPAR-H PSFs.
X1 X2 X3 X4 X5 X6 X7 X8
Influence of X uponY Available time Stress/Stressors Complexity Experience/Training Procedures Ergonomics/HMI Fitness for Duty Work Processes
Y1 1.0 Medium to high Medium to Medium Medium to Medium Low to medium Low to moderate
high high (complex
Available time (stress can (greater experience (poor layout can (illness, high (poor shift turn over of
or poorly
increase the High means that less time is result in increased levels of fatigue, information can reduce time
(ratio of time available conceived
time required to complexity required for actions and reaction time, or drug abuse available for crew response)
to time required) procedures
process can increase decisions; shifts the lessening the may increase
determine
information and the time margin in the available available time to time required to
how much
perform required) time in either direction) respond) decide or act)
time one
actions)
needs to act)
High (less 1.0 Medium to Medium Low to Low to medium Low Low
time may high medium (poor ergonomics
(more experienced (illness can
increase can contribute to
workers may experience (poor lower the
Y2 stress) increased
less stress) procedure threshold stress
workload and
Stress/Stressors completeness effects upon
physical and
or quality can performance)
mental stress)
increase
stress)
Medium to High 1.0 Medium to high Medium Medium Medium Medium
high (diminished
(stress can make (experience can mitigate (better (poor ergonomics (cumbersome work
capacity can
(little time the situation the effects of complex procedures can require more processes and supervision
result in simple
makes the task appear more decisions through reduce actions per task or can increase the complexity
Y3 situations
more complex b/c heuristics and actions) complexity) that the operator associated with maintaining
experienced as
Complexity complex, don’t perceive perform more equipment; increasing
complex or
simultaneous information) computations and uncertainty through poor or
overwhelming,
acts) calculations by miscommunication can
i.e., exceeding
hand or mentally) heighten complexity.)
channel
capacity)
Low Medium Low 1.0 Low Low Low Low
(affects ability (procedures
Y4 to recall can
Experience/Training information complement
acquired during the experience
training) level)

G-3
 X1 X2 X3 X4 X5 X6 X7 X8
Influence of X uponY Available time Stress/Stressors Complexity Experience/Training Procedures Ergonomics/HMI Fitness for Duty Work Processes
Low Low Medium Low 1.0 Low Low Medium
(ergonomics for (particularly important for
Y5
situations can procedure design review
Procedures make it difficult to and implementation))
follow
procedures)
Low Low Low to Low Low 1.0 Low Low
medium
Y6 (greater experience can
mitigate the effects of
Ergonomics/HMI
marginal ergonomics but
cannot override)
Low Medium to high Medium Low Low Low 1.0 Low to medium
(high
(poor ergonomics (there is some evidence that
complexity
such as lifting a poor safety culture can
Y7 may induce
requirements, can result in general lowering of
fatigue or
Fitness for duty interact with fitness for duty for an entire
amplify
medical work group).
circadian
conditions or
effects)
circadian effects)
Medium Medium Medium Medium Medium Low Low to medium 1.0
(multi-agent (procedures (illness and
complex can influence substance abuse
tasks require the or irregular
Y8 greater effectiveness work cycles can
coordination or occurrence affect crew
Work Processes ) of work dynamics and
processes) the effectiveness
of work
processes that
are in place)
* Relative relationship is defined as either low, medium or high.

G-4
 Appendix H
SPAR-H Development History

H-1
H-2
 Appendix H
SPAR-H Development History

Original Development (1994) HRA process, which could be improved upon as

state of the art in HRA improves.
Efforts directed toward development of SPAR-H
method focused upon producing a simple, general, The existing ASP HRA methodology was
and easy-to-apply method, which considers or developed in 1994 (Blackman and Byers, 1994)
accounts for actuation, recovery (to the extent that and, for clarity and convenience, is hereafter
it is present in the PRA model), and dependency referred to as the 1994 ASP HRA methodology.
through a consistent model of human behavior. A The enhanced and revised version of the
general criticism of HRA methods is the inability methodology, with minor exception, is the basis
to tie these methods back to first principles in for the SPAR-H 2004 revision presented in this
human behavior. Generally, methods identify a set report. The 1999 Version is referred to as the
of factors believed to be related to performance simplified plant analysis risk (SPAR) HRA
(e.g., stress and stressors, training, procedure method.
quality), or focus on classes of human error
(omission, commission, mistakes, slips), or even The 1994 ASP HRA methodology was developed
general classifications of human behavior (rule, to make an order of magnitude improvement in the
skill, and knowledge), and then manipulate those HRA practice of the accident sequence precursor
factors to arrive at a failure rate. The obvious (ASP) program (the previous method had been
problem with these approaches is completeness. limited to four human error probability (HEP)
How do we know that the set of identified factors values). The 1994 ASP HRA methodology made
is, in fact, complete? In developing SPAR-H, we use of a two-page worksheet to rate a series of
began with a model of human behavior and went performance shaping factors (PSFs) and
to operating events and behavioral sciences dependency factors to arrive at a screening level
literature to determine whether the model and its human error probability (HEP) for a given task.
associated elements covered the basics. Noteworthy features of the 1994 methodology
To our knowledge, no single HRA method begins were a derivation of PSFs from a psychological
with a theory of human behavior, to ensure that all model of human behavior, and an explicit
relevant factors are addressed and accounted for, dependency model. However, when compared to
and then works forward to identify demonstrated, the open literature and individual plant
underlying mechanisms that are known to examination (IPE) HRA data, the dynamic range
influence and be predictive of behavior. To avoid for HEPs in the 1994 ASP HRA methodology was
this basic flaw in method development, time was limited, and the taxonomy for distinguishing the
spent in identifying an underlying model of human processing (cognition) portion of a task from the
behavior from which to develop a clearly response (action) portion of the task proved
supportable and complete HRA method for SPAR. somewhat difficult for collaborators who were
non-human factors and HRA professionals to
Because there was a need for simplicity and apply. In addition, a more obvious link to human
usability, SPAR-H does not consider detail on a performance literature and human performance
finer level than it does. For example, more refined distributions was needed beyond the top-level
aspects of work process or of information model. This was addressed in the current version.
processing and decision making, such as situation
awareness and parallel versus serial search 1999 Revision
strategies, could have been brought into the model, The 1999 revision attempted to enhance the
but would have made it difficult to use and existing accident sequence precursor (ASP) human
interpret. We acknowledged that there was a need reliability analysis (HRA) methodology to make it
for method that was good enough to support the more accessible for the SPAR modeler to apply.

H-3
Although it may serve to support other modeling development was underway, there were 72
efforts and characterizations of human Revision 2QA models, which were made available
performance for the risk analyst, the primary for use by mid-1998 (Holahan et al. 1998).
function of the revised SPAR-H methodology will
still be to support the SPAR models. We believe Scope. The work to revise the ASP HRA method
that the SPAR-H method can serve other was cast as four subtasks:
functions, such as screening for most HRA 1. Review other current and emerging HRA
applications, and that when placed in appropriate methods for similarities and differences
logic modeling structures, SPAR-H can help
identify contributions to risk associated with 2. Adjust PSFs and/or influence weights based
human performance. However, readers are still on the review results and user comments
cautioned that this is a screening analysis tool and 3. Review and adjust dependency calculations
not meant to replace complete HRA methods. This based on the review results and user
being said, analysts must still apply a reasonable comments
standard of investigation and evaluation of
scenarios provided by PRA to obtain an accurate 4. Adjust base HEPs based on the review
analysis. results.

Simplified Plant Analysis Risk (SPAR) models The revision of the 1994 ASP HRA Methodology
have been developed by NRC for use in accident was completed in 1999 by the INL and remained
sequence precursor analyses for operating plants. in draft form. It was field-tested by NRC
These level 1 SPAR models are used to evaluate inspectors, SPAR model developers, and HRA
the estimated conditional core damage probability, analysts. Comments and experiences with the
given a specific initiating event or the existence of method were collected, and the method was
a specific condition at a plant. These models were addressed again in 2003 with expansion of the
developed initially as simplified models, i.e., screening method to LP/SD scenarios. This report
restricted number of initiating events (only those documents the latest version (2004 revision) of the
that were considered most common [transients and SPAR-H method.
loss of offsite power] or bounding for safety- 2004 Revision
related systems not challenged by the common
events); support systems when not modeled Uncertainty. The SPAR-H method as revised in
explicitly (only impact on frontline systems 1999 only determined point estimates for HEPs. It
modeled); and basic events rolled-up into super was desirable, for purposes of PRA, to develop a
components, resulting in smaller fault trees. method whereby uncertainties in the HEP
estimates could be propagated in the PRA.
Subsequent to the development of the first version Therefore, we set out to determine the uncertainty
of 75 plant-specific SPAR models, changes and distributions for SPAR HEPs.
additions to the models were identified and
implemented in Revision 2. Revision 2 models Distribution of HEPs. Previous approaches to
consist of the following: treatment of emergency HRA uncertainty. Since publication of THERP,
ac power was expanded; plant-specific features the lognormal distribution has become an accepted
impacting station blackout were added, in addition distribution for skilled performance. Sträter
to certain plant features identified in the licensee’s (2000) has added further weight to the argument
IPE submittals; and the BWR models were for using a lognormal distribution as set forth in
modified to include interdependencies among the THERP for HEPs.
power conversion and the condensate and feed
water systems. The models were revised to THERP postulates a lognormal probability density
accommodate the comments generated from a function (PDF) with a standard deviation of 0.42.
quality assurance review and are now designated A SD of 0.42 was obtained by assuming a 4:1
as the “Revision 2QA” SPAR models. Since some range ratio between the 95th and 5th percentiles for
plants have been shut down while model tasks performed under routine conditions.

H-4
However, it then goes on to say that the range and at-power conditions that was performed, it
ratios used in reliability analyses of NPP tasks are was determined that a separate worksheet for
considerably wider than the nominal 4:1 ratio; LP/SD should be developed. This worksheet is
thus, calling into question that approach. presented in Appendix B. Findings from field test
results of the SPAR-H method led to
Our review of the human performance literature improvements in the LP/SD and at-power
suggests that human performance may often worksheets. The following enhancements to earlier
follow either a normal log distribution, but that it versions of the worksheet were implemented:
also may follow a quadratic, or cubic distribution.
Also, the transformation from success space to 1. LP/SD Worksheet Enhancements
failure probability is not so straightforward. We
PSF time available for actions. The dynamic
believe that the mean value should be preserved.
range of influence for expansive time
Also, we advocate the use of a beta distribution to
available (50x nominal) was changed. The
model HGPs. Specifically, we use the constrained
range of effect for expansive time now uses a
noninformative distribution, which maximizes
multiplier of 0.01. When time available is
uncertainty about the mean HEP valve. This
determined to be 5x nominal, then a
distribution provides an adequate representation of
multiplier of 0.1 is used.
the upper bound and truncated at a value of one.
PSF Procedures. An additional
Analysts are therefore encouraged to use the CNI
level of influence was incorporated for
approach to uncertainty calculation discussed in
LP/SD. Incomplete or partial procedures
Section 2.6 of this report.
influence the base HEP by a factor of 20.
Sources of Uncertainty This influence is present for action and
diagnosis tasks. Analysts commented that
Unsurprisingly, the estimation of HEPs has this assignment would be potentially
uncertainty associated with it. It is obvious that our valuable for at-power conditions as well and
industry has done a much better job in collecting, this level has also been applied to the at-
collating, and analyzing equipment failure data power worksheets.
than human errors. As Swain and Guttmann
(NUREG/CR-1278 1983) point out, uncertainty in PSF time available for diagnosis. Time
HRA comes from such sources as: available multipliers were developed.
Diagnoses with extra available time
• Dearth of the type of human performance data available ranging from less than or equal to
useful to PRA/HRA 2x nominal are assigned a multiplier of 0.1;
expansive time (defined as >2x nominal) is
• Inexactness of models of human performance
assigned a range of effect from 0.1 to 0.01
• Inadequate identification of PSFs and their that may be assigned by the analyst. This
interactions and effects better reflects the increased uncertainty and
longer time horizons associated with a
• Analyst skill and knowledge limitations number of LP/SD tasks.
• Variability in performance (both within the
PSF Complexity. An additional level of
individual and between individuals).
influence for favorable complexity (i.e.,
All of the above, except the last, fall mainly into obvious diagnosis) was developed. The
the category of epistemic uncertainty. On the other multiplier associated for this category is 0.1.
hand, the innate variability in performance,
particularly within individuals, appears to be so 2. At-power Worksheet Enhancements
intractable that it may as well be regarded as PSF time available for actions. The
random. dynamic range of influence for expansive
time available was changed. The range of
Comparison of LP/SD and at-power conditions.
effect for expansive time now uses a
As a result of a qualitative comparison of LP/SD

H-5
multiplier of 0.01 and in addition to the use partial procedures influence the base HEP
of absolute minutes in earlier versions of by a factor of 20. This influence is present
SPAR-H, the relative time available in for action and diagnosis tasks.
conjunction with the time required for task
performance has been taken into account. PSF time available for diagnosis. The
influence of expansive time (>24 hours for
PSF Procedures. An additional level of diagnosis for at-power conditions) was
influence was incorporated. Incomplete or changed from a multiplier of 0.001 to 0.01.

H-6
 APPENDIX I
SPAR-H Review Comments

I-1
I-2
 APPENDIX I
SPAR-H Review Comments
Table I-1 on the following pages presents response to comments received during the peer review of the
revised SPAR-H method. The peer review followed an iterative format, in which comments were
solicited and received throughout the development of the document. After completion of the draft
NUREG report, a formal review period followed during May – October, 2003. The following
individuals participated as peer reviewers.

NRC ACRS PRA Subcommittee Buttonwood Consulting

Suzanne C. Black George Apostolakis Dennis Bley
James Bongarra Dana Powers
Mike Cheok Steve Rosen Scientech
Susan Cooper J. Sieber Jeff Julius
Dave DeSaulniers
Mike Franovich Sandia National Laboratory Brookhaven National
Claire Goodman John Forrester Laboratory
Hossein Hamzehee Hugh Whitehurst John O’Hara
Chris Hunter John Lehner
Paul Lewis EPRI
Erasmia Lois John Gaertner INL
Gareth Parry Nick Grantom Ron Boring
J. J. Persensky J. Grobbelaar Bruce Hallbert
Marie Pohida Frank Rahn Pat McCabe
Nathan Siu Gary Vine Jennifer Nadeau
Dave Trimble Dave Pack
Peter Wilson SAIC Marty Sattison
Alan Kolaczkowski April Whaley
ERIN
Steven Mays

A high-level summary of major changes made in response to the formal peer review comments follows:

• The scale the time available PSF was modified from absolute time intervals in minutes to time
intervals relative to the time necessary to complete a diagnosis or action.
• The definition of the stress PSF was broadened to include stressors, encompassing both psychological
and physiological components of stress that influence behavior.
• The definitions for diagnosis and action, as well as joint diagnosis and action tasks, were clarified in
order to make the SPAR-H worksheet usage more intuitive for analysts.
• Comparisons between SPAR-H and other HRA methods such as THERP, SHARP, and the ASME
standard were clarified.
• The SPAR-H worksheets now include an adjustment factor to ensure that the HEP cannot exceed 1.0.
• An additional PSF level for “insufficient information” was added to all SPAR-H worksheets. This
PSF level is weighted identically to the nominal PSF level, but it acknowledges circumstances when
information about PSF level assignment is unavailable.
• Where appropriate, levels were added to the PSF tables to account for the positive influence of certain
PSFs. It is assumed that in regulatory space, most PSF assignments will center around the negative
influence of PSFs.
• Throughout the report, numerous clerical errors and inconsistencies were corrected.

I-3
Table I-1. Formal peer review comments and responses.
Comment Source Response Scoping
1. The SPAR-H Method does not consider any credit for the opportunity to NRC Crediting recovery or failure to act is handled indirectly in the model. It is available See Section 2.8.
recover from an operator’s error that would be provided by additional in one of two ways: first, through explicit modeling on the part of the analyst and,
crew. second, by manipulation of PSFs.
2. The Time Available Nominal shaping factor criterion for power NRC A relative time scale with some time limitations is being adopted in response to See Section 2.4.4.1.
operation appears to use 20 minutes as a lower bound. (Also see similar questions. A minimum time is also presented because there is little evidence
Comments 8, 14, 30, and 78) that situations whose time is too compressed allow for recovery even if the action or
decision does not take long.
3. In the SPAR-H Method, no consideration is taken of the strength or NRC In a number of situations it is highly likely that cue strength is important. In SPAR- See Section 2.4.4.6.
quality of the information or cues available to the operator regarding the H, cue strength is evaluated in the assignment of the Ergonomics/HMI PSF.
necessary actions that must be taken to mitigate an accident initiator.
4. When one uses the SPAR-H Method worksheet, it appears that there are NRC Yes, the PSF levels are negatively skewed. It is expected that the analyst working in See Sections 2.4.3
more ways for the baseline probability for a Diagnosis task or an Action regulatory space will be reviewing events where the concern will be that more and 2.5.1.2.
task to be increased than there are for this probability to be reduced. influences are expected to be negative rather than positive.
5. The differences between the SPAR-H Method and other acceptable NRC SPAR-H is a simple worksheet driven HRA approach that is a composite method As resources permit,
HRA approaches such as EPRI or THERP are not clear. spanning several others. It was developed to support ASP models and has grown to detailed comparison
include support of the significance determination process (SDP), and to aid in other between SPAR-H
uncomplicated HRA efforts. At the time the method was developed details and EPRI methods
concerning the range of influence for PSFs and a detailed listing of PSFs were not could be conducted.
available. Hence, the SPAR-H tables do not make direct comparison with the EPRI
CBDT approach. While going to press, more information on the EPRI method was
available and information regarding that method has been incorporated in the latest
revision.
References to THERP and the comparisons with THERP are contained in various
sections within the body of the document. In general, SPAR-H offers the
advantages of unifying disparate approaches to HRA modeling while requiring the
analyst to consider a consistent, full range of PSFs and expands the diagnostic
approach found in THERP to include other considerations such as dependency on
subsequent events and PSFs.
6. The SPAR-H Method acknowledges the issue of interactions among the Sandia, Beyond determining an adjustment factor to assist in instances where double As resources permit,
PSFs but offers no explicit way of handling the interactions and SAIC counting among highly related factors is moderately possible, no utilization of a could be performed
assumes the PSFs can be multiplied as independent influences. Perhaps matrix or other process is currently intended. General guidance is that analysts be at a later date.
simplified guidance and a worksheet could be derived that would allow sensitive to the possibility of double counting and make their PSF assignments
the user to define strong relationships among the PSFs. accordingly. The suggested adjustment factor is located in a calculation space
provided in the SPAR-H worksheets.
7. Several issues regarding the appropriate assignment of the nominal rate Sandia, Guidance on finding extremely high or low reliability situations is related to Out of scope.
should be addressed: SAIC declaration of context. In SPAR-H, this is done bottom-up through the assignment
(a) Guidance as to how to define extreme failure or high of the PSF values.
reliability situations may be appropriate.
(b) There should be a system that ensures consistent linking of There are no current plans to develop additional PSFs or to develop linking of
learned factors to the assigned PSFs. learned factors acquired through event analysis to the assigned PSFs. This could be
achieved through establishing a SPAR-H user group or by expanding the scope of
the current SPAR user group.

I-4
 Comment Source Response Scoping
8. The definitions of the various PSFs do not seem appropriate in all cases. Sandia, The PSFs are anticipated to be applicable to the majority of events and situations to See Section 2.4.4.1.
For example, assigning “barely adequate” as less than 20 minutes fails SAIC be evaluated but probably not to all. If the time available is less than 30 minutes and
to recognize that for some actions, 1 or 2 minutes may be “lots of time,” not thought to influence performance beyond the normally expected range or if the
and for another action, even one hour may be barely adequate. We analyst determines that there is an insufficient amount of information to make an
recommend re-examining the eight PSFs to ensure that they reflect the appropriate assignment, he or she does not have to assign a negative value to time
dimension appropriately. As part of this, consider “relative” definitions. available.
Also, consider tradeoffs such as whether recent, situation specific
simulator training can be more important than whether an operator has For actions, the time PSF has been modified to relative time multiples, which is the
less or more than six months of experience. (Also see Comments 2, 14,
ratio of time available to the time required.
30, and 78)
9. Some PSFs, e.g., fitness for duty and ergonomics/HMI, should be Sandia, Fitness for duty and ergonomics/HMI are more likely to be used in performing See Sections 2.4.4.6
examined only under special scenarios. Additional discussion as to SAIC retrospective analysis where such information may be available from comments and 2.4.4.7.
when to address such factors would be useful. determined from the AIT report. Review of behavioral sciences literature and events
determined in NUREG/CR-5763 support the use of these factors. Prospectively, the
analyst would need access to drawings, have performed walkdowns, reviewed near-
miss databases, etc. before assuming ergonomics to contribute beyond the nominal
case. Instructions currently acknowledge that for model development the generic
nominal value can be used and that the PSF assignment be re-evaluated when
applied to a plant specific case.
10. It is recommended that the dimensions related to dependency be re- Sandia, The dimensions in the dependency model could be re-evaluated at a later time. Out of scope.
evaluated to be sure that they are the best choices (given the expected SAIC Currently, these are offered as a refinement to some basic work by Swain et al.
applications of SPAR-H) for post-initiator actions. (NUREG/CR-1278, -4772) and have been applied in the SPAR-H process since
1999.
11. Check for consistency and cross-referencing of section numbering. Sandia, This suggestion has been implemented. Throughout report.
SAIC,
EPRI
12. Consider deleting Section 2.7.2. The data from the studies cited seem Sandia, The material from this section is included because there has been no externally See Section 2.7.2.
very far removed from the PSFs in the SPAR-H method and their SAIC distributed version of SPAR-H prior to this report and we think that some
multipliers. background in these studies should be provided.
13. Looking at the Shutdown SPAR LP&SD event trees, based on the NRC The decision whether to remove operator diagnosis from all LP/SD trees for PWR See Sections 3.2 and
SPAR-H definition of diagnosis, the top event (“Operator diagnoses the loss of inventory, loss of offsite power trees, and loss of RHR trees is beyond the 3.3.
event”) should be removed from all the PWR Loss of Inventory Trees, scope of the present report and is worthy of review by the SPAR User Group and
the Loss of Offsite Power trees, and the Loss of RHR trees. PRA analysts. The necessity of the diagnosis basic event may prove to be a function
of the context of the scenario including existing plant complexity factors,
availability of indication, and the specific plant involved.
14. Using multiples of nominal time to define available time categories Sandia This approach is implemented with minor modification. Also the analyst has the See Section 2.4.4.1.
seems preferable to using actual time intervals. (Also see Comments 2, opportunity to address the potential for overly conservative assignment by use of the
8, 30, and 78) “Obvious diagnosis” category available with Procedures.

I-5
 Comment Source Response Scoping
15. The definition of stress used is extremely broad, covering both Sandia We have broadened the definition of stress to include “stressors.” The analyst may See Section 2.4.4.2.
psychological and physiological stress. The stress that is likely to select an appropriate level of experience if he or she, upon review, believes that
degrade performance, especially during an emergency, is acute individual differences in stress are likely to appear as a function of experience and
psychological stress, but individual differences, especially experience, that he or she has enough data to determine the level of experience for rotating
affect the degree of performance degradation. One approach is to crews.
change the name of the PSF from “Stress” to “Stressors.” We then
acknowledge not knowing if particular individuals experience
psychological stress.
16. Regarding experience and training, the high multiplier of “3” might be Sandia The argument is intuitively appealing. However, the range of influence for training Out of scope.
correct under normal conditions. However, under emergency conditions, and experience in SPAR-H mirrors that used in other first generation methods. The
especially conditions that are life threatening, it seems that a higher application of SPAR-H for use in risk/vulnerability studies where such factors as
multiplier is needed. heat, fire, explosion, debris, loss of communications, and inadequacy of existing
procedures may figure into personnel response and plant risk was not part of the
current project scope.
17. An operator who is fit for duty under normal conditions might not be fit Sandia Agreed. Again this application extends the boundaries of the method. We would Out of scope.
for duty in life-threatening emergencies. Under such circumstances, suggest using the appropriate level of fitness for duty for this context and explaining
good fitness means that uncommon physical exertion will not degrade the basis of the assignment in the mandatory comments section of the worksheet.
performance. This is an area for potential further development. (Also see Comment 16.)
18. A number of corrections are necessary. NRC, (a) Corrected Throughout report.
(a) A correct reference to IEEE Standard 1082 should not EPRI (b) Corrected
include a “P” designation. (c) Corrected
(b) The introductory References section is missing the 1999 (d) Corrected
Apostolakis citation. (e) The inconsistency has been corrected
(c) IEEE Standard 1574 should be cited as “P 1574” as it is a (f) The Table has been modified to be consistent with Appendix G. The
draft document under development. diagram is consistent with a path diagram where bi-directionality of
(d) In Appendix C, page C-10, the dependency condition table effects is modeled, rather than an influence diagram where the
indicates that the dependency for the task failure is “low.” directionality is typically one way.
However, in the calculation of task failure probability with
formal dependence at the bottom of the page, the value for
“moderate dependence” is used.
(e) In Appendix C, page C-10 (and elsewhere), the
nomenclature for Probability without Formal Dependence
(Pw/od) and that for Probability with Formal Dependence
(Pw/od) appear to be incorrect. The same nomenclature is
used to describe two difference dependencies.
(f) In Appendix G, Table G-1, the relationship as depicted in the
influence diagram between “stress” and “fitness for duty”
does not appear to be in agreement with the explanation/
relationship level provided in Table G-1.

I-6
 Comment Source Response Scoping
19. The report would benefit from being reorganized; the current version is NRC Different organizational schemes for the document have been under discussion. Out of scope.
confusing in that the theoretical basis and the comparisons with other However, it was determined that it would be better to have users understand the
methods are intermingled with the description of the method. It would technical basis of the method by forcing them to review the history and issues
be more usable if the report were rewritten with the main body associated with the method prior to learning the mechanics of the method. It is
providing a description of the method, and the detailed discussion of the believed that this will help ensure more consistent application by users.
human behavior model, the comparison between the models, and other
historical material moved to Appendices.
20. In the Executive Summary, the first sentence in the second paragraph is NRC Corrected. See Executive
misleading. The SPAR-H Method is not “a method for predicting Summary.
human error,” but a method for estimation of the probabilities (HEPs) of
the human failure events included in the SPAR PRA models.
21. It is stated that the SPAR-H Method can be applied at the task or sub- NRC Decomposition is dictated by the approach of the HRA analyst to the PRA and by Answered but
task level. The definitions of task and subtask suggest that a task is the circumstances surrounding what is being modeled. We do not, a priori, suggest benchmarking
modeled at the level of a basic event of the SPAR Model, which may which level of decomposition is most appropriate for all or even specific situations. against other HRA
include many sub-tasks. Since the method is based on adjusting the base This is a challenge for HRA in general. When inter-analyst comparisons are made, methods or among
case HEP, it is difficult to see how this can be performed in a consistent even second-generation HRAs may present different values and inconsistencies in analysts is desirable
manner if the analyst can choose to use the SPAR-H method for tasks or what is considered, including the level of decomposition. Finally, it is thought by the but out of scope.
subtasks within the same PRA. Different analysts may use different authors that the degree of difference among analysts will be constrained by the use
although internally self-consistent approaches; inconsistencies will of the worksheets, which forces consistent consideration of the same PSFs,
emerge when comparing HRAs. dependency, and approach to uncertainty. A benchmarking exercise would benefit
the field.
22. It would be better to refer to SHARP1 rather than the original SHARP NRC We are in the process of obtaining access through the NRC PM to SHARP1 (EPRI Out of scope.
as a framework for performing HRA. The revised SHARP1 method 1999) from the EPRI website. At the time that this report went to press, SHARP1
discusses the identification and definition of the human failure events in was not yet available for review. We therefore acknowledge the importance of the
a more systematic and complete way than did SHARP. reference but show only limited discussion of particular aspects of SHARP1 within
the body of this report.
23. In the Executive Summary, the discussion in the first paragraph of the NRC Additional guidance differentiating diagnosis from action has been added to the See Executive
section entitled “Overview” is confusing, particularly in differentiating definitions portion of the report, and this information will be brought forward to the Summary.
between diagnosis and action. executive summary.
24. The characterization of the SPAR-H method in Section 1.1 is NRC Yes, SPAR-H is a quantification tool. Accident sequence model development does See Section 1.1.
overstated. SPAR-H is a quantification tool. It is not structured to help help identify HRA events or sub events. In prospective analysis, the HRA approach
identify the human failure events that should be included in the PRA can help to identify errors of commission that are then fed back into the PRA
model; this is a function performed when developing the accident process. In retrospective analysis, the same is true; HRA sub events not identified in
sequence models. the plant PRA but which had an impact upon risk can be identified, modeled, and
quantified and the PRA model updated accordingly.
25. Since SPAR-H is primarily a quantification tool, it is not necessary for NRC The user audience is anticipated as being diverse. Any HRA including SPAR-H See Section 4.2.
this document to discuss different HRA approaches for identifying the should not be performed in a vacuum. Although not a part of SPAR, per se,
human failure events (HFEs) for which the tool is intended to provide presenting recognized search strategies does not, in our opinion, take away from the
probabilities (HEPs). So, for example, the part of Section 4.2 that document. It is hoped that readers will access these reference materials in greater
discusses the ATHEANA approach to HRA is superfluous. detail. We did not develop our own strategy because there are several good ones out
there including ATHEANA and EPRI approaches.

I-7
 Comment Source Response Scoping
26. What is meant by diagnosis is not altogether clear. With the NRC The definition for diagnosis has been clarified. Most of the plant models have See Section 2.2.1.
preponderance of symptom based procedures, there are very few real diagnosis events for which HEPs are required, and it would be imprudent not to do
diagnosis events. Rather, the concern is more whether the symptoms so. Also a number of actions observed during operating events have indicated errors
have been observed and interpreted correctly to transition to the in planning and other cognitively demanding tasks, still others involved improper
appropriate set of instructions. prioritization of activities within the corrective action backlog, or improper resource
allocation. The analyst can model entry into the correct procedures and would
probably make use of the ergonomics (i.e., cue strength, systems feedback,
simultaneity of faults), training and experience, procedures, and work processes in
determining the appropriate HEP.
27. In the case of continuous action pages, the operator may be required to NRC This situation appears to involve a combined or joint diagnosis and action HEP, See Section 2.2.3.
monitor several indications while taking actions. This is a significant instructions for which are present on the worksheet. Analysts would look closely at
degree of cognitive activity. Would this be modeled as diagnosis rather time available, complexity, training, and stress for these activities.
than action? What is the guidance?
28. Mistakes can have a significant common cause failure potential. When NRC Intuitively we believe that mistakes can have a significant common cause effect. We See Section 2.3.
considering dependency in Section 2.3, SPAR-H provides no guidance are less certain regarding the extent to which other failure types can contribute to
on how to determine whether an error is a slip, lapse, or a mistake. common cause. The determination of the potential error type by the analyst is part of
Therefore, this discussion is superfluous, or needs to be expanded upon. the HRA search process and PRA scenario development and not part of
quantification per se. That is why we mention its importance. The details of how to
discover and model these various error types are to be found in other references. The
point to be made is that SPAR-H can be applied to any or all of these potential error
types.
29. Currently, there are two base case HEPs, one for diagnosis and the other NRC Our approach at this point in time has been to address the different diagnosis states See Sections 2.2 and
for action. Does it make sense to consider two different cases for by PSF assignment using procedures and complexity and not to distinguish between 2.3.
diagnosis, one for the case where the crew is in a knowledge-based diagnosis differences for knowledge-based versus rule-based, i.e., procedures,
mode and another for the procedure following model? An alternative domains. Some PSF levels such as “obvious diagnosis” could, conceivably, apply to
would be to handle such differences through one of the PSFs, e.g., either situation. We would expect more diagnoses in LP/SD to be in the knowledge
complexity. If this is the case, it should be clearly discussed in Section realm and some subset of those diagnoses to be associated with different degrees of
2.4.5. complexity. Research could help to determine the extent to which given the same
scenario, some crews could be operating in a mode (i.e., skill, knowledge, rule)
different than would be other crews.
30. It’s not clear that a nominal time can be defined in terms of a specified NRC A modification to levels of time available has been determined and is contained in See Section 2.4.4.1.
number of minutes. The nominal time should be related to the strength the worksheets and in the definitions portion of this report. The present approach
of the signature of the event, and the degree to which the operator is assumes that the role of training and strength of the signature of the event (i.e., cue
trained to recognize it. (Also see Comments 2, 8, 14, and 78) strength or systems feedback) can be represented as a PSF influence.
31. It is likely that PRA analysts will choose a level of stress based on some NRC We have broadened the definition of stress to include stressors, which may make it Out of scope.
other PSF, such as the time available. Since this is already accounted easier for analysts to make stress level assignments. We agree and have assumed
for, it would mean double counting. The levels of stress need to be that analysts would help to determine the level of stress based on their understanding
established taking into account the information on the context that can of the context of the situation as opposed to one or two other PSFs. When situations
be gleaned from the SPAR PRA models. are highly negative, we would expect a number of PSFs to be negative as well.
Double counting is a concern, and we attempt to address the PSF inter-relationship
issue elsewhere in this report but believe more research is needed. Finally, we have
included an adjustment factor to the worksheet to reduce the potential effects of
double counting on HEP estimates.

I-8
 Comment Source Response Scoping
32. In Section 2.4.4.3 about the Complexity PSF, should there be separate NRC On a plant specific basis, we would expect the analyst to make use of the “obvious See Section 2.4.4.3.
discussions of the diagnosis and action? Some actions, e.g., pump diagnosis” category in order to reduce conservatisms used in employing the nominal
recirculation, are relatively easy to diagnose but require a very focused rate. We agree with the reviewer, that it is possible to have a joint HEP where the
activity. task is complex but the diagnosis portion of the HEP is less complex. Hence the
reason for separate worksheets when constituting the joint HEP.
33. The section on Experience and Training seems to suggest that it is the NRC Frequency of training is now factored into the definition of levels of experience and See Section 2.4.4.4.
crew experience and level of training that is most important. However, training PSFs. To the extent that the HRA analyst has knowledge of the crews’
the level, frequency, and type of training may vary from scenario to scenario-specific training this information should be factored in the training and
scenario, and this might be a more important factor when assessing the experience .PSF level rating.
relative HEPs for the different HFEs.
34. Examples for the definition for Nominal Ergonomics for shutdown in NRC This addition has been implemented and can now be found in Section 2.4.4.6. Note: See Section 2.4.4.6.
Section 2.4.4.6 should include the availability of RCS level In general, we are concerned about placing too much information in the general
instrumentation and RHR system instrumentation for BWRs. For instructions because the scope can easily become too large and exceptions to the
PWRs, the PSF definition of Nominal Ergonomics should include the examples can routinely be found. It is safest for the SPAR-H analyst to list the
availability of RHR system instrumentation, the availability of RCS available instrumentation when s/he indicates a nominal rating on the worksheet.
temperature instrumentation, and the availability of RCS level We have implemented documentation for nominal assignments as a requirement.
instrumentation.
35. In Section 2.4.4.8, the issues related to Work Processes in the control NRC In documenting the method, attention to detailed analysis regarding work processes Out of scope.
room, e.g., inter-crew communications could be different than those associated with maintenance activities has not received great emphasis. We agree
relevant to in-maintenance activities. Expanding the examples in this that development of additional work process examples relative to work processes
section, where possible, is essential. would be beneficial. This could include informing the control regarding status of
planned or current in the field activities as well as lock out tag out status of
equipment. See 34 above.
36. The dependency between HFEs is driven by the context of the accident NRC Yes, dependency between human failure subevents can be driven by the accident See Section 2.6.
sequence. While it is not necessary to take account of the positive sequence. It could also be driven by continued use of a flawed procedure and other
dependency effect on the probability, two failures separated by a considerations. We agree. In general, separation of the two failures by a success
success may strengthen the case for independence between these HFEs. would strengthen the argument for independence. There may, however, be instances
where this is not the case.
37. Should diagnosis and action failures be modeled separately? This is NRC This gets to the question of whether or not to use a combined diagnosis/action HEP See Sections 2.2 and
largely a question for the system analyst. If the impact of both failures is that is allowed on the worksheet and is a situation specific decision. When there is 2.6.
the same, there is no need to represent them separately. When, however, serial diagnosis, we agree with the reviewer. Separate the terms, and evaluate the
there is a potential dependency between the diagnosis of two serial second diagnosis for dependency influence.
events, then it may make sense to separate the terms.
38. When calculating the HEP, should the success in diagnosis be NRC Currently, we are somewhat conservative in estimating the HEP. We do not add in See Sections 2.4.3.
accounted for when adding in the probability of failure to perform the the success.
action?
39. The last paragraph of Section 2.7.1 is incomprehensible without having NRC The section has been clarified. See Section 2.7.1.
a copy of THERP open to the right page. More background information
is necessary.
40. The purpose of Section 2.7 is not clear. In some sections, the focus NRC The purpose of 2.7 is to describe how SPAR-H fits into a PRA using Bayesian See Sections 2.7.1
seems to be on the representation of uncertainty; in others, it is more methods. Part of this discussion must, therefore, center on the how and why a and 2.7.6.
related to the shift in the mean as a result of the strength of a PSF. certainty uncertainty approach is used in lieu of others. Also, since SPAR-H uses
PSFs as thresholds to justify movement of an expected value, this discussion is also
applicable to this section.

I-9
 Comment Source Response Scoping
41. Sections 2.7.1, 2.7.6, and 2.7.7 seem to fit more naturally in Section 2.6, NRC We have opted to break out dependency and uncertainty discussions separately See Section 2.7.
since they are strongly related to the discussion of the uncertainty because we believe them to be worthy of equal consideration.
distribution.
42. The focus in Sections 2.7.2 – 2.7.4 seems to be on demonstrating that NRC The reviewer is correct; the section does present the range of influence of PSFs on Out of scope.
the impact of certain PSFs on human performance can be represented by human performance. PSFs need to be represented in the method, and there is a
specific distributions. This suggests that they are related to the strength quantitative relationship among PSFs and performance. The issue of the uncertainty
of the PSF, which is not uncertainty as used in this report, but it regarding a PSF or combinations of PSFs is not addressed separately from the
represents a variability in human performance as a direct result of a uncertainty surrounding the mean HEP. PSF uncertainty is not well researched in
PSF. HRA but is a reasonable future topic.
43. Section 2.7.5 seems to undermine the model of independent PSFs. NRC This argument has been made by different reviewers. We agree that it is by Out of scope.
Figure 2-4 may be an honest assessment of the complexity of the convention that we in the HRA community (as evidenced by a number of methods,
problem, but it just reinforces this message. SPAR-H included) view PSFs as independent when performing calculations.
Currently, the burden is on the analyst to be vigilant in preventing double counting.
Research in this area is highly desirable. Appendix G lists some of our ideas
regarding potential relationships that need to be determined by review of operating
events data. Finally, we provide an adjustment factor on the worksheets to help
reduce the influence or potential for double counting.
44. Section 2.8 reads as if the authors are suggesting that all recovery NRC Not all recovery actions must be explicitly modeled. The analyst can also use work See Section 2.8.
actions, even those resulting from a second person checking in practices for a second checker, if the facility routinely uses second checkers to good
preinitiating events, should be modeled explicitly. This would advantage. Decomposition and representation will be a function of the scenario and
unnecessarily complicate the model for HFEs that are typically not the way in which the analyst has accounted for human performance in the rest of the
significant contributors. HRA.
45. In Section 3.1, there is a subsection called “validation of PSFs against NRC We agree that validation implies actions other than those carried out by the authors. See Section 3.1.
operating events.” What does this mean? It is clearly not a calibration of The sentence in question has been reworded as “reviewed against operating events.”
the factors in any numerical sense. In NUREG/CR-5763, human performance influences in high profile events subject
to ASP analysis were identified. These influences can be mapped to the PSFs
present in SPAR-H.

I-10
 Comment Source Response Scoping
46. The comparison of SPAR-H with the ASME Standard should be NRC Table 4.5.5-2(d) of the ASME Standard refers to the assessment of the probabilities See Section 5.2.
focused on Tables 4.5.5-2(d) and 4.5.5-2(g), since these are the tables of the preinitiator human failure events and calls that they be addressed via a
that address the quantification method. systematic process. It notes that THERP and ASEP are acceptable methods. SPAR
nominal rates, PSFs, and range of effects have been calibrated against THERP and
ASEP and should meet the criteria. The approach does call out differences among
screening and detailed assessments. SPAR-H is a simple HRA method and does not
differentiate between screening and detailed analysis; all actions or diagnoses are
subject to the same procedure. We have no equivalent to ASME category I. We call
for the analyst to note the quality of written procedures or administrative controls,
and the quality of HMI, which are category II and III items for ASME. We are less
detailed on establishing the maximum credit that can be given for multiple recovery
opportunities. SPAR-H assesses the potential for recovery of preinitiator or post
maintenance or post calibration tests through the use of PSFs such as the work
process PSF. Independent verification using written check-off lists, work shift or
daily checks, and other factors should be noted in the comments field of the work
sheets. Otherwise the nominal HEP is thought to encompass recovery as a function
of self check or check by others. If there is reason, then the analyst may explicitly
model and quantify the recovery using logic structures and SPAR-H values. SPAR
does allow for dependency and providing point estimates as called forth in this table.
Finally, the analyst is required to check the reasonableness of HEPs in light of the
plant’s history procedures, operational practices, and experience.

SPAR-H suggests that analysts always refer to operational history to help inform
and check HEPs. This is consistent with guidance suggested by ASME.
47. In the example in Appendix C, it would be helpful if the information NRC This information was excerpted from another report. We agree that at least a listing See Appendix C.
required for the analysis were to be summarized as an example of the of the type of information necessary to support the analysis should be included.
sort of documentation needed to apply the method. What is given is Recent events preclude a detailed listing of this information, but the point is well
essentially a summary of the event tree structure. Some important taken. Although we are constrained by the need not to release sensitive information,
information that is missing includes: (i) an estimate of the time window we have added a generic event tree to further clarify the existing example.
for making the diagnosis, (ii) the information required to recognize an
SGTR, and (iii) the procedural guidance that helps to lead to the
diagnosis.
48. The examples of PSF characterization in Appendix G are for events. An NRC The PSFs are mapped to operating events as opposed to the actual base models. See Appendix G.
equivalent for HFEs in a base case model might be helpful. There is some mapping of PSFs to base case models in NUREG/CR-6753,
Appendix C.
49. An issue that seems to be overlooked is a qualitative screening process EPRI SPAR-H purposefully, with the intent of forcing consistency, requires that the same See Section 2.4.4.
to identify which PSFs might be important for a particular human error set of PSFs always be applied (evaluated) for every HEP under consideration. PSFs
evaluation, and if any PSFs that are not defined will dominate. It that the analyst feels might dominate will have to be mapped to existing PSF
appears that the SPAR-H method assumes that the fixed list of eight structure and basis for this explained in the comments column. We do believe the
PSFs is sufficient. Ideally, a way of estimating the impact of non- list of eight is sufficient for a simplified HRA approach.
modeled PSFs should be addressed.

I-11
 Comment Source Response Scoping
50. The authors should fully address the relationship between the input EPRI, SPAR-H, when used in a PRA model, utilizes a “Bayesian” setting to represent Out of scope.
factors and the conversion to probabilities within this document, rather NRC uncertainty. When this approach is applied, the application should be able to defend
than relying on the internal processing of SAPHIRE to carry out this any modeling assumptions used during the analysis. Ad hoc methods such as
step. SAPHIRE automatically converts the raw inputs into the mean classical statistics via regression analysis or arbitrarily selecting distributions such as
value of a constrained non-informative (CNI) distribution, which keeps log logistic are not defensible in the context of SPAR-H. Consequently, the INL, in
all values between 0 and 1.0. The basis for selection of the CNI defining the uncertainty, focused solely on the facts that (a) the HEP is bounded
distribution versus alternate methods (e.g., log logistic) should be between 0 and 1, (b) we believe that the result from the SPAR-H worksheet
further explored since it is not provided in the text. represents an expected value, and (c) we really do not know how diffuse the HEP
distribution is for specific events. Using these three assumptions leads one to some
type of maximum entropy distribution for the uncertainty. Coupling this general
distribution type with the fact we believed we know the posterior expected value
then directly implies use of the constrained noninformative (CNI) distribution as
discussed in the text. Of course if evidence specific to certain human events provide
information related to the plausibility of that event, formal hierarchal Bayesian
methods may be used, but this type of detailed uncertainty analysis is beyond the
scope of SPAR-H. Further, the use of the CNI distribution does not limit
applications solely to the SAPHIRE software. For example, the text discusses the
beta function call provided by the EXCEL software and how to use it to represent
the CNI distribution. All modern PRA software will, therefore, have the ability to
utilize the CNI distribution.
51. The NRC should consider the use of logistic regression analysis as part NRC The use of logistic regression analysis in lieu or in addition to the current approach Out of scope.
of the current approach is not within the current project scope. It would require a separate research effort.
52. In Section 2.7.2, the authors have selected a very interesting set of basic EPRI More detailed review and linking of the performance distributions to CNI Out of scope.
human performance research topics labeled as distributions. The distribution parameters would be worthwhile. We think that development of a set of
discussions appear to have some relationship to the SPAR-H method. rules for setting the feasibility of an action under specific accident scenarios could
The document could be improved by more clearly linking the basic be explored. We are less certain that either of these activities will substantially
research models to the assignment of the CNI distribution parameters, contribute to a simplified HRA approach, since it is likely that the relationships will
and perhaps a set of rules for setting the feasibility of an action under need to be considered in a numerical hierarchical Bayesian framework.
specific accident scenarios.
53. In Section 3, the authors tied the error rate to the ranges used in other EPRI This is an interesting suggestion for future improvements. The measurable data Out of scope.
methods. An improvement would be to relate the process of selecting a would have to be defined along with baseline tasks. We argue to keep the current list
calibration factor to some form of statistically measurable data. In of PSFs the same.
taking this step some baseline tasks should be defined, which can be
matched to the current and future PSFs that might be analyzed.
54. A major weakness in the SPAR-H process is relating the assignment of EPRI The PSF is tied to the HEP as opposed to the functional objective. In obtaining the Out of scope.
PSF values to a defined functional objective typically analyzed in a functional objective, multiple actions with different individual PSF qualities could
PRA. There is little guidance in defining the context of the action or be involved. Which PSF fits best is scenario and plant specific and we would not try
relating the type of PSF to where it best fits in the PRA. to predefine this for multiple applications. In the SPAR-H approach, the context is
referenced through the assignment of PSF levels as opposed to declaring extreme
error forcing context or some other type of assignment. This determination would
have to be the subject of further study.

I-12
 Comment Source Response Scoping
55. The fitness for duty PSF would seem to be a characteristic related more EPRI In event analysis, or in support of SDP, making changes to the uncertainty Out of scope.
to the responsibility of the individual than of all the crews, and for this distribution as opposed to keeping things on a PSF level would make the method
reason would be more appropriately modeled as a contributor to the overly complex. Fitness for duty may or may not be an individual factor. For
uncertainty distribution rather than to the mean value of the HEP. example, consider a situation where the entire crew has pulled a double shift or has
responded to an emergency requiring a strong physical effort during that same shift,
etc.
56. In the case of PSFs for timing, the authors attempted to simplify the EPRI Time is present in terms of generic transient response for groups of scenarios and Out of scope.
issues by addressing timing in the same way as other PSFs. This can then this time is used in HRA assessments, allocated either to diagnosis or action or
cause some difficulty in integrating with the PRA models, since timing to both as stated by the reviewer. Viewing or constructing sets of simultaneous
is generally addressed with a generic transient response for groups of equations, one for HEP and one for timing, might refine things but would certainly
scenarios. The HRA assessments then use the timing information to diverge from the spirit of keeping SPAR-H a usable HRA approach.
address the details of the response, which involves allocating time to
both the cognitive and action elements or to the other elements
analyzed. It seems more appropriate to view the HEP evaluation with
simultaneous equations, one for the HEP and one for the timing.
57. The review and evaluation of the method in Section 5 is a very EPRI Agreed. The self evaluation could be expanded and method would benefit by Out of scope.
interesting self-evaluation, however it should be subjected to subjecting it to independent benchmarking.
independent benchmarking before drawing any strong conclusions
about its capability to meet the criteria in various standards.
58. Based on the worksheets and idea that inspectors can fill out the forms, EPRI SPAR and SPAR-H training is under development. Multiple PRA courses are Out of scope.
it would seem that they would need some specialized training in the offered to NRC staff including the various NRC regions.
following areas: (i) PRA training on the meaning of accident sequences
and role of human actions to be able to define a context for the action
(HEP) or integrated result (HFE); (ii) training on the basic elements of
HRA and the relationship to the human factors evaluations.
59. The worksheets in Appendices A – E address the results as probabilities, EPRI The worksheet now contains an adjustment factor that precludes HEPs >1, where the See Section 2.7.6
where the body of the report indicates that the results are some kind of HEP is a best estimate of the mean that is used as an index when input into the CNI. and Appendices A-E.
an index that is used as input to the CNI distribution to prevent Formerly, the SPAR-H method stated in the case of very high HEPs that the HEP
probabilities greater than 1.0. The worksheets can produce probabilities be set directly to 1.0. This heuristic treatment is the same as that for PSFs such as
greater than 1.0, which violate the fundamental probability theory “fitness for duty” wherein if the operator is unfit for duty, or if there is simply not
axioms. enough time to perform the task, the HEP is 1.0 regardless of any of the other PSFs.
The inclusion of an adjustment factor does allow the analyst to produce sufficiently
negative HEPs where warranted while staying within the acceptable range for the
beta distribution.
60. Since the values generated in the worksheets are inputs to the EPRI Strictly speaking the PSF levels are scalars and the mean HEP is a probability. There Out of scope.
dependency analysis, the dependency results become some mixture of is a precedent in THERP and other methods for multiplying them. In THERP there
probabilities times indexes, which lose meaning with regard to is also precedent for modifying an HEP that has been adjusted for PSFs through the
distributions and axioms of probability. assignment of levels of dependency (ranging from 0 to complete dependency). In
SPAR-H PSFs are multiplied by the mean HEP as if independent. We are aware of
this simplified approach creating some difficulties. By highlighting dependency, the
analyst forces discussion and hopefully improves the PRA. We argue that a higher
HEP can be expected in situations where the there is a linkage to previous failed
subevent(s) and that this is a direct means by which to call attention to this fact. For
additional insights on the uncertainty treatment, see Comment 50.

I-13
 Comment Source Response Scoping
61. The combination of the diagnosis and action probabilities should follow EPRI We believe that they do. SPAR does not, however, assign a dependency rating for See Section 2.2.3.
the basic rules of logic. individual components of a subevent. For example, a task that has diagnosis and
action sub components can be treated as a combined failure where either is sufficient
to fail the task. A dependency calculation is not performed within the same task.
Some reviewers have argued for and against. We do not multiply these two
components together simply because it is not an AND (i.e., intersection) gate
situation (1E-2 x 1E-3) – for the action to fail, we only need diagnosis to fail OR the
action to fail. Either failing to diagnose or failing to take the appropriate action is
sufficient in a combined HEP to fail. If the task can be completed without diagnosis,
then the diagnosis contribution does not contribute to the failure space represented
by the fault tree where the task is modeled and, consequently, would not be factored
into the model nor the SPAR-H worksheet.
62. Why have different sheets for full power and low power operations EPRI (a) If it would be beneficial there could be a worksheet check off box that See Section 2.4.4.1.
when only one number is modified slightly? identifies pre- from post initiator activities. We employ a basic human
(a) It would seem more reasonable to address preinitiators and performance model. The people don’t change, only the PSFs under those
post-initiators in separate sheets. This could be as described conditions, even on a routine basis. SPAR-H would allow for diagnosis
in the ASME standard and others that recognize a significant component of maintenance trouble shooting to be accounted for if the analyst
difference between diagnosis (cognition) and actions so desired.
(execution) and between a response to an event and routine (b) Both the data and the processing are contained on the same sheet to make
test and maintenance tasks. things easier for analysts in the field. This has been viewed as a positive
(b) The current sheets appear to mix the data and processing for feature through user feedback.
these types of events. (c) The PSF types are expected to remain the same; problems in mapping to the
(c) If the PSF data set of the PSF types vary, then new eight included PSFs can be noted by the analyst in the comment fields. We do
worksheets should be developed that reflect the research not know whether application to extreme events or external events would
modeling differences between the types of tasks. require new PSFs or new dynamic ranges.
63. Why have different sheets for diagnosis and action when the PSFs are EPRI Using separate sheets accords better human factors, because it reduces the memory See Sections 2.2.1 -
the same and the variation in the inputs is nearly the same? burden on the person doing the assessment. The same is true for remembering the 2.2.3.
different PSF levels as a function of the assignment of diagnosis or action, which is
best not left to memory. Data could be collected to evaluate the point further.
64. Why address timing independently within an HEP evaluation three EPRI Timing is addressed in this way, because there may be instances where time is Out of scope.
separate times (i.e., available time for diagnosis, action, and dependency relevant for diagnosis but not action , i.e., it could be nominal and there are instances
analysis)? where time does or does not contribute significantly to dependency. If the question
is merely an organizational issue, the form could have time addressed in one area of
the form; however, there are enough levels to the PSFs that this probably would not
reduce work or confusion.

I-14
 Comment Source Response Scoping
65. Primary failure modes for diagnosis (mistakes in cue interpretation, etc.) EPRI Identifying failure modes themselves (assignment of slips, lapses, or mistakes) is not Out of scope.
and actions (slips, errors in keeping track of procedural steps, HMI, etc.) made in the worksheet per se. The SPAR-H approach makes it the analyst’s
the evaluation should address these issues in a more appropriate way. obligation to declare such information in part of the HRA write up or embed such
information as part of the error description or as part of PSF information.
(a) Time dependent response models could be considered. Determining which error mode is more likely is part of the HRA search process and
PRA scenario development effort.
(a) Development of time dependent response models or a series of such
(b) Consider the systematic linking of simultaneous equations to models would serve to complicate the analysis and reliability inherent in
evaluate time as a resource limited PSF case. applying the worksheets. The models would have to be developed and
evaluated against a variety of conditions and scenarios.
(b) Equation development and validation is not part of the scope of the
present effort.
66. Nomenclature issue. SPAR-H does not take advantage of the ASME EPRI The language among HRA methods is not consistent. The INL has used HEPs, See Appendices A –
definitions for the assessment. HEPs are single evaluations of a nominal HEPS and conditional HEPs to represent differences in the level of E.
diagnosis/action before the dependency assessment. HFEs are integrated analysis, consideration of PSFs, and dependency. For example, unsafe acts are used
assessments made after the dependency assessment that represents the by some analysts and can be the result of single or multiple HEPs. Only a portion of
overall failure to manage the sequence for post initiator actions. unsafe acts end up as HFEs. And the mapping of UAs to HFEs can be one to one or
many to many. The adoption of a common language and resolution of differences in
terminology is a worthwhile goal.
Errors with subscripts were noted.

The errors with worksheet subscripts have been noted and corrected.
67. It is a good idea to address the dependency when two separate tasks are EPRI The dependency calculation is typically used for an AND gate situation. However, See Appendix D.
involved in the same accident sequence. The application of Swain’s we could conceive of how poor procedures could imply a propensity for a failure
dependency formulations appears to be misapplied. The way that human that wasn’t present and inhibit good diagnosis (HEP1), call out an incorrect action
actions appear in the same accident sequence is through AND gates in (HEP2), and failure on a second action (HEP3) is now increased because of the
the same model. If they go through OR gates, then they appear in difficulty caused by the failure on the first action. There is a linkage between 1 and
separate sequences and no dependency assessment is needed. Recheck 2, and a common linkage among all three by virtue of the procedure’s influence on
the formulation of dependency in Appendix D and some aspects of performance. The dependency table on the worksheets has been modified to make
calculation. determinations easier and more explicit.
68. What is missing from Appendix F is a comparison to the base case EPRI Such a comparison is not part of the analysis performed in support of documenting Out of scope.
conditions that are assigned to the mean HEP value. This would be the method for this report.
needed to illustrate what to look for in evaluating a PSF change.
69. Provide an intra-dependency PSF matrix whenever multiple PSFs are EPRI There have been separate suggestions along these lines. However, the work has not Out of scope.
applied. been performed on how to implement such a matrix, qualitatively or quantitatively.
As such, it remains beyond the scope of this report.
70. Existing industry methods such as EPRI CDBTM (EPRI TR-100259) EPRI This has been adjusted in the body of the report. See Section 2.4.
and THERP (NUREG-1278) are not recognized as complete HRA
methods.

I-15
 Comment Source Response Scoping
71. The SPAR-H process does not specify how HFEs consisting of multiple EPRI SPAR-H suggests that the HFEs be decomposed to the level of tasks as appropriate. Out of scope.
tasks are to be addressed. Some applications are performed at a higher level than others. The main guidance is
that the analyst be consistent with the level of decomposition. It may be possible to
develop rules regarding the number of subtasks or tasks that could or should be
combined. This could be an area for further research. In general, determining the
correct level of decomposition is not a criticism or challenge unique to SPAR-H.
Some analysts are more holistic and others atomistic. If an analyst does use rules for
combining sub tasks or sub events, s/he should make explicit those rules as part of
the HRA analysis effort.
72. Table 2-3 provides THERP multipliers, which are not generally EPRI Yes, these multipliers were determined from our review of THERP and an See Table 2-3.
presented in THERP as multipliers. This should be indicated in the text. annotation has been added to the report.
73. The interpretation that THERP provides a multiplier of 50 in the EPRI In part, the reviewer’s comment assumes that HEP for action tasks where there is an See Table 2-3.
absence of procedures is incorrect. THERP provides an HEP of 0.05 for absence of procedures should be 0.05 and suggests that a diagnosis task’s HEP =
“when written procedures are available and should be used but are not 0.05 should also exist. In that situation, the latter has a multiplier of 50 and the
used.” The multiplier is inferred if assuming a base HEP of 1E-03, former a multiplier of 5. That has not been our approach.
which is the Action HEP in SPAR-H. However, if the Diagnosis HEP of
1E-02 in SPAR-H is assumed, then this multiplier would only be 5. The reviewer is correct that 1.0E-3 with a factor of 50 yields 0.05, and this
interpretation can be traced back to NUREG-1278 and actually applies to errors
regarding the omission of an item when procedures are not used.

However, our interpretation of Swain for the diagnosis case is that diagnosis in the
absence of procedures yields an HEP of 0.5 (50 x 1E-2 = 0.5), hence the multiplier
of 50 still holds. This is our interpretation; Swain provides no guidance for a
diagnosis situation in the absence of procedures.

We reviewed ASEP, and in Tables 7.1 and 8.1 it appears that actions without
procedures have an HEP = 1.0. We prefer the THERP version. Diagnosis in ASEP
uses time available calculation for diagnosis then adjusted for procedures. For
diagnostic situations without the written procedures, the HEP is assigned the value
HEP = 1.0 and then, according to the guidance, the HRA is stopped. In our
judgment, ASEP appears overly conservative. The use of the multiplier of 50 that
we selected was an engineering judgment on our part that softened the available
ASEP value.
74. SPAR-H considers “time available” for diagnosis and action separately. EPRI This is a good point, and we have incorporated this into the instructions. See Section 2.4.4.
The time available is usually obtained from physical transient data and
applies to diagnosis and action. Will it be obvious to the user that
diagnosis time will decrease action time and vice versa?

I-16
 Comment Source Response Scoping
75. The SPAR-H method discounts the lognormal distribution on the basis EPRI Our use of the beta distribution was an attempt to increase the precision of the math See Section 2.7.6.
that it could produce an HEP > 1, and it adopts the beta distribution and produce means and upper and lower bounds that would easily support PRA
instead. The beta distribution may be an equally valid approach, but the calculations available in software workstations such as SAPHIRE. This addresses a
lognormal is widely used in industry, not only in HRA but for number of complaints regarding not only SPAR-H but other methods employing
component failure rates as well. lognormal distribution of error (e.g., THERP, etc). Additionally, some of the work
on human performance in the behavioral sciences literature suggests that many other
distributions are better descriptors of human behavior than is the lognormal. For
additional clarification on why the INL did not use ad hoc methods such as
lognormal distributions see Comment 50.
76. The use of Fitness for Duty as a PSF may be onerous. Based on EPRI This concern may be true for prospective analysis, which in the use of this PSF may See Section 2.4.4.7.
discussion, this would, for example, pertain to different shifts having be more difficult. However, in SPAR event analysis, this variable affords the analyst
different failure rates. This could imply that the HRA needs to take into the opportunity to account for fitness of the crew as it figured in the event. Usually,
account different shifts. Given the inherent uncertainty in HRA, the one or two crews at most are involved, and data are present.
additional effort to model shifts would not be justified by the value
added.
77. The SPAR-H report discusses several issues of timing. There is a EPRI This is an interesting discussion. If the analyst has strong evidence to believe that See Section 2.4.4.1.
systematic problem in that the PSF is set on the time available to the crew has inaccurate or imprecise knowledge regarding the time available before
complete an action before some irreversible, undesirable event occurs to events cannot be corrected, this should be recognized. It is difficult to determine the
the reactor. This time window is used to set the stress PSF for the operator’s internal state. It could be due to training/experience or misleading
operator. The problem is that, for this PSF, the operator is not always indication, e.g., tank level. We would adjust these two PSFs as opposed to time
aware of the time available to diagnose or perform an action. The available. The rate would be based upon the time available (its ratio to the time
operator will have a perceived time window, which may or may not be required) and these other PSFs.
consistent with reality.
78. For the diagnosis PSF, the first timing demarcation is set at 20 minutes EPRI The timing demarcation was interpreted from THERP, which is the basis. However, See Section 2.4.4.2.
and described as “barely adequate.” The basis for selection of 20 we have taken this comment to heart; the PSF assignment now takes into account
minutes is not indicated or justified. Note, having 22 minutes available two factors; relative time and absolute time. We do reduce the HEP for “at-power”
rather than 20 minutes, changes the PSF from 10 to 1. (Also see situations where there is extra or expansive time available, beyond a certain
Comments 2, 8, 14, and 30) minimum corresponding to interpretation of the THERP diagnosis curves.

The data that are available supports the notion of thresholds. Where such
information has not been available we have relied upon relative times as a means
of better approximating the expected HEP.
79. Table 3-5 is a comparison of the differences between power operation EPRI The table is based upon input from a variety of sources including individuals See Table 3-5.
and shutdown operation. It is biased from the perspective of a PRA familiar with LP/SD situations. Potential uses of systems such as RWST gravity
analyst who is unfamiliar with shutdown PRA and more comfortable drain, spent fuel pool cooling, reflux boiling and fill and spill will be considered as
with power PRA. The table is used to change the method of PSF potential future modifications to the table in question. The authors will also make
assignment for shutdown PRA. The comment here is that differences changes that decrease the likelihood that users will be led to perceive LP/SD as
between power and shutdown operation are not substantial differences abnormal as opposed to less frequently performed evolutions.
and do not form a basis for changing the PSF scheme.
80. The SPAR-H method does not provide sufficiently detailed criteria for EPRI In 1999 some inter-rater reliability work with the method was performed with Out of scope.
the HRA analyst to select the appropriate category of PSF quality. This positive results. In 1999 and in 2000 additional user comments were employed to
will lead to different analysts assigning different PSF levels and getting enhance the definitions. We see value in obtaining reliability ratings beyond the
different answers, with no way to resolve the differences. national lab and NRC HQ as a means of further pedigree for the method.

I-17
 Comment Source Response Scoping
81. The NRC should work to establish reliability measures for SPAR-H ACRS The idea to extend reliability measures to the community at large is a good one. Out of scope.
beyond the national laboratory and a limited number of staff. However, this program has always operated on iterative feedback. All users have
been and continue to be encouraged to make the program aware of problems in
applying the procedures or any aspects of the method. This has been the approach
taken for a number of years and has proved to be quite valuable in improving the
method and standardizing the current approach.
82. Why does the NRC need more than one HRA method? ACRS SPAR-H is a simplified HRA method that is intended for use in quick turnaround See Section 2.4.1.
event analysis, building plant models, and by inspectors to support the NRC
significance determination process (SDP) process. SPAR-H can highlight issues or
problems that then can be addressed by more labor intensive approaches such as
NRC ATHEANA or others in industry that have a recognized approach in
determining highly specific contexts in which the likelihood for errors of
commission is heightened. In forcing consistency among analysts with limited
training, SPAR-H uses nominal rates to account for a mixture of omissions and
commissions. It also uses a basic human performance model that does not
differentiate between latent and active human failure events, leaving that to the
analyst’s error description and PSF evaluation(s). SPAR-H is also not geared to
identify unique error causal mechanisms as is the case with some second-generation
HRA methods.
83. Why is there a fixed number of PSFs when other methods have a ACRS We believe that the eight PSFs that SPAR-H uses in quantification will cover the See Sections 2.4.1
different number or in some cases seem to have an unlimited number majority of the situations that the analyst building plant models or performing event and 2.4.3.
depending upon what an analyst may discover during the HRA process. analysis will need. The list is fixed to force consistency in the PSFs considered by
analysts every time the method is used. We also believe that most of the PSFs that
other methods may use that are not present in SPAR-H can be mapped to existing
PSFs. For example, corrective action program failures, or design inadequacies can
be mapped to the SPAR-H work processes PSF.
84. Are PSFs assumed to be independent? Are we double counting ACRS PSFs are not independent. The authors have included cautionary statements in the See Section 2.5.
somewhere? For example, stress can be related to the time available; in text, per the suggestion to sensitize users. The adjustment factor can reduce double
others the impact of the task can be so overwhelming that having extra counting to some extent but it is based upon mathematics as opposed to theory why
time may not really reduce the stress associated with task performance. some PSFs might be more highly related than others. Finally, in Appendix G there
INL should consider having a warning or at least sensitizing the user to are some findings regarding the relationship among PSFs, and the Discussion
the potential for overlap among PSFs. Section of the report calls for the need for the HRA community to provide research
in this area.
85. It appears that the message from second generation HRA is that context ACRS Generally speaking, the message from second generation HRA is that context See Section 5.2.
matters. Do you agree, and is this represented in some way in SPAR-H? matters. The assignment of context requires some level of expert judgment. Context
in SPAR-H is acknowledged through the assignment of PSF levels for diagnosis and
action task types and through dependency assignment. SPAR- H does not focus on
context-driven search schemes for error identification but instead provides some
references for the less trained analyst where he or she can go to support the error
identification and modeling process.
86. There is a fundamental difference between the nominal condition or NRC Agreed. An additional PSF level for insufficient information has been added to the See Appendices A –
having insufficient information, making it difficult for the analyst to Staff worksheets. We agree that different types of uncertainty and/or degrees of E.
know whether the nominal or some other PSF level applies. uncertainty exist between these situations.
Additionally, in terms of uncertainty, it would appear that these are two
different situations as well.

I-18
 Comment Source Response Scoping
87. The method and presentation would benefit if more examples of NRC The current report documents the current state of method development including the Out of scope.
application, especially those involving multiple PSFs and simultaneous calculation approach and definition of terms. We agree that it would be good to
conditions could be developed and reviewed in the report. Particularly, develop additional examples to help ensure consistency and understanding among
more examples regarding PSF assignment during LP/SD would be of analysts and applications of the method. The extent to which simultaneous
benefit to analysts. conditions change the interaction among PSFs and performance is a challenge to the
field of HRA.
88. Unlike other PSF situations evaluated by the analyst where positive INL SPAR-H does not account for the influence of positive dependence. The same is See Section 2.4.4.2.
influences can reduce the base HEP, there seems to be no credit given or true for the positive effects of stress. While it is factual that there are data that some
allowance made for situations where an analyst may wish to assign a level of stress is associated with positive performance, the SPAR-H analyst does not
positive influence associated with stress. This seems to be in evaluate the influence of stress in this way. As a beginning point, we assume
disagreement with information presented previously, i.e., in the INL (unless otherwise indicated by a negative fitness for duty selection) that we are
review of human performance distributions, where there seems to be dealing with a reasonably alert and motivated operator. This assumes a positive
evidence that stress can be a motivating force and that optimum influence for stress and arousal that is accounted for in the nominal value for stress.
performance is associated with a certain degree of stress or arousal. For our purposes, the stress and stressors supporting positive operator or crew
performance are assumed to be addressed when the analyst selects the nominal
stress category.
89. The current instructions associated with symptom based procedures are INL We agree. If the symptom based procedure is found to be inaccurate or awkwardly See Section 2.4.4.5.
disturbing in that poor symptom based procedures would still reduce the constructed, then the procedures PSF should be negatively rated. Section 2.4.4.7 has
nominal HEP. Please review. been modified to incorporate this suggestion.

I-19
NRC Form 335 U.S. NUCLEAR REGULATORY COMMISSION 1. REPORT NUMBER
(2-89) (Assigned by NRC. Add Vol., Supp., Rev.,
NRCM 1102, and Addendum Numbers, if any.)
3201,3202
BIBLIOGRAPHIC DATA SHEET
(See instructions on the reverse) NUREG/CR-6883
INL/EXT-05-99509

2. TITLE AND SUBTITLE

THE SPAR-H HUMAN RELIABILITY ANALYSIS METHOD

3. DATE REPORT PUBLISHED
MONTH YEAR
August 2005
4. FIN OR GRANT NUMBER
W6355
5. AUTHOR(S) 6. TYPE OF REPORT
D.I. Gertman, H.S. Blackman, J.L. Marble, J.C. Byers, C.L. Smith Technical

7. PERIOD COVERED (inclusive Dates)

10/1999 – 8/2003

8. PERFORMING ORGANIZATION B NAME AND ADDRESS (If NRC, provide Division, Office or Region, U.S. Nuclear Regulatory Commission, and mailing address; if contractor, provide
name and mailing address.)

Idaho National Laboratory

P.O. Box 1625
Idaho Falls, Idaho 83415
9. SPONSORING ORGANIZATION B NAME AND ADDRESS (If NRC, type "Same as above"; if contractor, provide NRC Division, Office or Region, U.S. Nuclear Regulator Commission,
and mailing address.)
Office of Nuclear Regulatory Research
U.S. Nuclear Regulatory Commission
Washington, DC 20555
10. SUPPLEMENTARY NOTES
P. O’Reilly, NRC Program Manager
11. ABSTRACT (200 words or less)

In support of the Accident Sequence Precursor Program, (ASP), the U.S. Nuclear Regulatory
Commission (NRC) in conjunction with the Idaho National Laboratory (INL), in 1994 developed the
Accident Sequence Precursor Standardized Plant Analysis Risk Model (ASP/SPAR) human reliability
analysis (HRA) method, which was used in the development of nuclear power plant (NPP) models. Based
on experience gained in field-testing, this method was updated in 1999 and renamed SPAR-H, for
Standardized Plant Analysis Risk-Human Reliability Analysis method. In 2003, to enhance the general
utility of the SPAR-H method, and to make it more widely available, the method was updated and
reviewed for its applicability to low-power and shutdown applications. This document presents the
current version of the SPAR-H method, along with guidance, definitions, improvements in representing
uncertainty, and increased detail regarding dependency assignment for HEP calculations. This report also
contains comparisons between this and other contemporary HRA approaches and findings specific to
application of the method to low power and shutdown events.

12. KEY WORDS/DESCRIPTORS (List words or phrases that will assist researchers in locating the report.) 13. AVAILABILITY STATEMENT
Unlimited
Human reliability analysis (HRA), probabilistic risk assessment (PRA), standardized plant analysis risk
(SPAR), human performance, accident sequence precursor (ASP)
14. SECURITY CLASSIFICATION

(This Page)
Unclassified
(This Report)
Unclassified
15. NUMBER OF PAGES

16. PRICE

NRC FORM 335 (2-89)