See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235643628

PRACTICAL TOOLS FOR TASK ANALYSIS

Conference Paper · April 2009

CITATION READS
1 6,279

1 author:

Jacques V Hugo
Semi-retired
41 PUBLICATIONS 137 CITATIONS

SEE PROFILE

All content following this page was uploaded by Jacques V Hugo on 03 June 2014.

The user has requested enhancement of the downloaded file.

Sixth American Nuclear Society International Topical Meeting on Nuclear Plant Instrumentation, Control, and Human-Machine Interface Technologies
NPIC&HMIT 2009, Knoxville, Tennessee, April 5-9, 2009, on CD-ROM, American Nuclear Society, LaGrange Park, IL (2009)

PRACTICAL TOOLS FOR TASK ANALYSIS

Jacques Hugo

Pebble Bed Modular Reactor (Pty) Ltd

1267 Gordon Hood Avenue, Centurion 0046, South Africa
jacques.hugo@pbmr.co.za

ABSTRACT

For most human factors practitioners, hierarchical task analysis (HTA) is right at the center of
their activities. Because of its ability to be adapted to a wide range of situations and needs it is
believed to be almost a “universal task analysis method”. In spite of this, HTA is regarded as
difficult, even by experienced practitioners. Practical problems, such as lack consistency of format
and presentation, often make the results difficult to interpret. There have been attempts to develop
software tools to support the analyst and also to achieve some order in format and presentation. In
spite of many years of development, very few tools have emerged that have achieved these goals.
The few tools that do exist have extensive functionality for producing graphical and textual reports,
but because they are all special purpose tools, they all lack the flexibility of a general purpose tool.
The paper describes how a general-purpose tool like Mindjet MindManager® is used at PBMR for
the representational as well as the analytical aspects of HTA, for example, capturing task
information from various source documents directly into hierarchical formats and then analyzing
and extending that information further to represent detailed task information. The paper also
discusses the rationale for deviating from the conventional ways of visual representation associated
with HTA.
Key Words: Hierarchical Task Analysis, Software tools, HTA

1. INTRODUCTION

1.1 Task Analysis in the Nuclear Industry

Task analysis has long been accepted as a key process in the design, development or
upgrading of nuclear power plant operations. Regulatory bodies in all countries regard task
analysis as mandatory, because it can describe the role of humans in safe plant operations and the
mitigation of emergencies and adverse conditions.
Much of the general acceptance of the importance of task analysis stems from lessons
learned from past incidents, such as Three Mile Island and Chernobyl, but also from a more
mature understanding of operator abilities and limitations and their behavior under operational,
emergency and maintenance conditions.
Hierarchical Task Analysis (HTA) is just one of the techniques regularly applied to ensure
that our knowledge of human performance and behavior is used in the rational allocation of
functions to humans or machines and that the tasks assigned to humans are described accurately.
It is also used to inform the design of workplaces, human-system interfaces, work practices,
training programmes and operating procedures.

Page 1 of 14
 Jacques V Hugo

A range of tools (both paper and computer-based) have been developed for various human
factors engineering activities, including requirements analysis, task analysis, human reliability
analysis, prototyping and human performance modeling. Most of these tools have been developed
in-house by the users and are not commercially available, or are aimed at a specific domain, such
as defense, maritime, or aviation (see for example http://carlow.com/hsitools.html). Those that are
commercially available, such as MicroSaint, Task Architect and ConcurTaskTrees, are special-
purpose tools that in most cases do not offer everything an analyst needs for all the complexities
of task analysis.
This lack of tools has lead to many practitioners using labor-intensive tools for HTA, like
paper-based forms, Excel spreadsheets and even PowerPoint and Visio.
Practical experience at PBMR has shown that Mindjet® MindManager®1 offers potential for
representing HTA information in ways that might have been “underused” by human factors
practitioners. MindManager is normally associated with concept- or mind-mapping, but its true
strength lies in its capability to handle large hierarchical information structures. Because it is also
a general-purpose information mapping tool, it can be used to accommodate most of the
requirements for an HTA support tool. A later section will describe how MindManager is used at
PBMR to capture task information from various source documents directly into hierarchical
formats and then manipulated further to represent detailed task information.

1.2 Assumptions and limitations

It is assumed that the reader is familiar with the basic principles of task analysis and this
paper therefore does not attempt to describe the underlying theory and process. For a
comprehensive discussion on task analysis and related topics, the reader is referred to the sources
listed in par. 7
The paper focuses on the use of a particular software tool and does not describe or compare
the characteristics and functionality of existing software tools like TaskArchitect, Euterpe and
ConcurTaskTrees.
It must also be emphasized that, as a generic human factors technique, there is no “standard”
way to perform Task Analysis and the methods and practice may differ significantly among
organizations. The discussion and examples in the paper are based on the practice at the Pebble
Bed Modular Reactor company (PBMR), but a detailed “how-to” description of the process and
the step-by-step use of MindManager and other software are beyond the scope of this paper.
Paragraph 2 below only provides an overview of the process.

2. OVERVIEW OF THE PBMR TASK ANALYSIS PROCESS

2.1 Objectives of the Task Analysis process

In PBMR, as in the rest of the Nuclear Industry, the objective of the task analysis process is
to ensure that all human tasks that are needed to accomplish the safe and efficient execution of
plant functions (process control, monitoring, maintenance, diagnosis, calibration and testing, and
event recovery) are identified, analyzed and described in sufficient detail to determine the design
requirements and assess risk. The analysis also allows the organization to compare the demands

1
Mindjet Corporation. (www.mindjet.com)

Page 2 of 14
 Jacques V Hugo

that the system makes on the operator, with the capabilities of the operator and if necessary, to
alter those demands, thereby reducing the probability of error and achieving successful
performance.
The PBMR Task Analysis Process was also designed to:
• Provide one of the bases for making design decisions for Human-System Interfaces (HSIs)
and other systems that require human involvement.
• Form the basis for specifying the requirements for the interaction with system components,
special tools and other maintenance equipment.
• Assure that human performance requirements imposed by the design do not exceed human
capabilities.
• Provide basic input for developing operating and maintenance procedures.
• Provide basic information for defining the staffing, training and communication requirements
of the plant.
• Form the basis for specifying the requirements for the displays, data processing and controls
needed in the control room to carry out tasks in a safe and efficient manner.

2.2 Task Analysis Phases

The process begins with an analysis of the operational demands inherent in the plant design.
These operational demands are represented in high-level operational goals, which are then
associated with, or shown as being dependent upon, certain functions that must be performed by
specific systems (as identified during the Functional Analysis process). Some of these system
functions may require human intervention for various reasons, while others may be automated to
some extent. These allocation decisions are not taken arbitrarily, but are governed by strict rules
that form part of the Function Allocation process and the Task Analysis methodology. In the final
analysis, all operator tasks must relate to system functions and to operational goals.
As described below, task analysis is conducted over three phases in increasing level of detail
to match the maturity level and the evolving nature of system design information.
2.2.1 Task Analysis Phase 1
During the first phase, only high-level task information is captured from System Operating
Descriptions (SODs) that describe the high-level operational characteristics of systems, as well as
the conceptual role of humans in the plant.
The analysis comprises a preliminary function-based identification of the likely role of
operators. The analysis produces a list (prioritized as far as possible) of systems for further
analysis and an indication of the level of breakdown required in the further analysis phases. The
results are entered directly into MindManager as a set of high-level task trees with systems
matched to primary operator roles, role characteristics, estimated task criticality and priority
rankings. (see Figure 5).
2.2.2 Task Analysis Phase 2
This phase comprises a hierarchical breakdown of operator tasks per system, linked to the
existing plant Functional Breakdown. The preliminary function allocation of Phase 1 is reviewed
and confirmed. The results of this phase include:

Page 3 of 14
 Jacques V Hugo

• Detailed Hierarchical Task Analysis (HTA) worksheets and HTA charts, compiled per
system and operational scenario obtained from the SODs. This is where MindManager’s
“map markers” and detailed annotations are used extensively (see Figure 6 and Figure 9).
• Operational Sequence Analysis (OSA), using the Sequence Diagram Editor2 tool. This links
operator tasks to operational sequence information obtained from SODs. This representation
is then provided to those groups in the organization responsible for operating procedure
development and operator training.
• Worksheets consisting of task listings and human-system interaction requirements. These are
required for the design of high- and low-level human-system interface screens for the
relevant system and the plant overall. At this point, the original MindManager information
can be exported to Excel and Word for further refinement.
2.2.3 Task Analysis Phase 3
This comprises a detailed analysis of tasks and operational sequences for critical tasks (i.e.
those contributing most to plant safety), as identified in the first phase. Tasks are verified against
the Functional Breakdown as well as the System Breakdown and plant operational goals. Later
iterations and detailed analyses may integrate operating procedures where indicated by task
criticality. The final iteration consists primarily of using the IPME3 software for simulation and
analysis of operational sequences as well as the analysis of human performance and workload for
the safety critical tasks identified in Phase 1. The final results are compiled in summary format
for inclusion in a Human Factors Analysis Report for each system.

3. REQUIREMENTS FOR AN HTA SUPPORT TOOL

3.1 Presentation conventions

In addition to its basis in cognitive psychology, a large part of HTA has to do with the visual
presentation of task information. Current practice is to a large extent influenced by the visual
conventions associated with HTA, as described by, for example, Kirwan and Ainsworth (1992),
Annett (in Diaper and Stanton, 2004), Hollnagel (in Salvendy, 2006) and Lehto and Buck (2008).
These conventions have become well established in the human factors community and there is
little evidence of attempts to investigate the usability of such conventions, with or without
dedicated tools.
One of the most enduring conventions is the shape of the HTA chart. The most common
shape of the HTA is a vertical hierarchy, represented visually in horizontal layers, with the lowest
level tasks at the bottom. There is however no good reason why a hierarchy (in task analysis or
anywhere else) cannot be represented vertically (that is, a tree lying on its side). On paper this is
often a more practical solution because it utilizes the space of the common portrait layout more
efficiently and makes it unnecessary for the reader to turn the page horizontally.
The following figure illustrates two ways of representing a hierarchy vertically.

2
Effexis Software. (www.effexis.com)
3
IPME - Integrated Performance Modelling Environment. A task network and discrete-event modelling tool, based
on MicroSaint, from Alion Science and Technology, MA&D Operation, Boulder, CO.

Page 4 of 14
 Jacques V Hugo

Figure 1: Alternative vertical layouts

Since this appears to be a matter of taste for most analysts, it should therefore be offered as
an option by HTA support software.
Similar conventions apply to the way other HTA information is represented, for example
plans and resources. In the following section, alternative treatments of these elements will also be
described.

3.2 Why the analyst needs help

With reference to the practice in the armed forces, Hone and Stanton (2004) considered that
"The most popular tool for use when conducting an HTA appears to be Microsoft Excel - and this
for representational purpose only". They also mention other tools like Microsoft PowerPoint and
OrgChart (organizational charting software) and mention that such tools do not offer any
assistance to the analyst in the conduct of the HTA.
Our experience at PBMR confirmed that “trivial” analyses could indeed be done in Excel
spreadsheets, or even in outline form in Word. However, most people would probably agree that
task analysis for a new nuclear power plant is a non-trivial undertaking and that the analyst needs
all the help he can get. It is certainly not cost effective to simply add more analysts when the task
is extensive. This is where tools are needed. Ainsworth (personal communication, September
2008) stressed that the main purpose of a tool should be to enable the analyst to really understand
the nature and structure of a task. If the tool requires the user to attend to the complexities and
functionality of the software itself, it will detract from the task analysis process and this may lead
to errors and omissions. Thus, any task analysis software should assist the users by forcing them

Page 5 of 14
 Jacques V Hugo

to think about the relationship between different tasks, without overburdening them with needless
complexity.
Hone and Stanton (2004) believe that a software tool to support HTA would need the
following main characteristics:
1. It should support the development of the sub-goal hierarchy and plans in the three different
formats of HTA representation (that is, tabular, graphical and outline).
2. Enable editing and verification of the analysis to percolate through each of the
representations.
3. Support extended analysis of the sub-goal hierarchy.
4. Enable further extensions of the analysis to be added.

While these objectives appear to be valid for the task analysis process overall, a systematic
approach makes a clear distinction between activities as they change over the task analysis
continuum: data collection, organization, analysis, modeling and simulation. Some of these
activities require complex judgement, while others may be regarded as almost “mechanical”. The
data collection and organization phases require a tool that helps the analyst to understand and
manipulate the structure of the raw data and to identify where additional data need to be
collected, whereas the analysis and modeling phases are more dependent on a deep understanding
of the concepts, principles and rules underlying the tasks being analyzed. The last phase,
simulation, is normally only performed for complex tasks, and here the analyst is dependent on
specialized tools like IPME.

4. USING THE MINDMANAGER® SOFTWARE FOR HTA

4.1 The power of visual thinking

The main purpose of charting and graphing tools is to make abstract data concrete, thereby
helping people to see, explore, and understand large amounts of information. Furthermore, a tool
that interactively engages the user in information visualization helps to:
• reduce the burden of searching for data in a large information space;
• enhance pattern recognition, whereby information is organized in graphical space by its
relationships that are otherwise more difficult to recognize;
• provide a manipulable medium that, unlike static diagrams, enables the exploration of the
information space.
• increase cognitive resources, because it becomes a visual resource to expand human working
memory.
A graphical tool thus effectively becomes “a tool to think with”. This is exactly the strength
of “mind mapping” and “concept mapping” tools like MindManager, PersonalBrain4,
MindMapper5 and Smart Ideas6.

4
TheBrain Technologies, LP. (www.thebrain.com)
5
Simtech Systems, Inc. (www.mindmapper.com)
6
Smart Technologies, ULC (www2.smarttech.com).

Page 6 of 14
 Jacques V Hugo

4.2 Thinking with MindManager

While no tool can replace the analyst’s insight, experience and skill, one should not
underestimate the value of a tool that makes information structures and relationships visible.
Because of this, Hone and Stanton (2004) emphasize that an effective software tool must make
provision for specific representation, interaction and usability features. The following table is an
item-by-item analysis of these proposed features and how they are handled by MindManager:
Table 1: Support tool features required
Item Description How this is supported in
MindManager
1. Task levels It should be able to accommodate an adequate number No limit, but rather influenced
of task levels by the visual size of the chart
2. Plans It must be possible to specify and show Plans in the User definable
various graphical and textual representations
3. Sub-goal hierarchy Break down sub-goal tasks to an appropriate level No limit - see item 1
4. Notes/Annotations Per goal, sub-goal, task and Plans, as general Extensive and user-definable
comments and annotations
5. Zoom Change the level of detail and size of display 10 - 400%
6. Auto-update Any change to an item should be automatically Extensive, and user-definable
reflected throughout the hierarchy.
7. Formats Provide graphical (horizontal, vertical, tree), Diagram - horizontal, vertical,
Collapsible outline and Table (import and export) tree;
formats. Outline - collapsible;
Table (Export to Excel)
8. File menus Offer easy access to files, modules and various objects Windows standard
imported into the HTA or exported from the analysis
9. User knowledge Capturing of input from subject matter experts should Can be imported from source
be easy and effective documents
10. Task Allocation Provide indication of allocation of tasks to resources User definable
11. Auto task Provide configurable, hierarchical numbering of tasks All levels
numbering
12. Mark thread as Indicate which tasks are terminal tasks User definable
stopped
13. Split screen view Show more than one view of the analysis Split horizontal or vertical views
simultaneously (e.g. graphical and text view). of tree
14. Color coding Provide configurable coding for levels, categories, User definable
attributes, plans, etc.
15. Macros Provide methods to automate certain frequent actions User definable
16. Templates Allow users to use standardized display and reporting User definable
conventions.

This is an obvious oversimplification of how MindManager may be used for HTA. As

pointed out in par. 4.1, visualization is essentially a cognitive activity; therefore, using a graphical
tool for HTA is a combination of thinking about structures and meaning, and manipulation of
objects. The graphical structure serves as an anchor for the detail information found in the
underlying task descriptions and in this way it helps to give structure to the analyst’s mental
model of the task domain.
Following this “thinking and doing” approach, we find that MindManager offers extensive
features for manipulation and representation:

Page 7 of 14
 Jacques V Hugo

• MindManager is ideally suited to the hierarchical breakdown of work domains and

associated tasks to as many levels as necessary (but typically not more than 6 or 7, which is
practical for most HTA analyses). The analyst must still decide on the breadth and depth of
the analysis. The software doesn’t help with this decision, but will support whatever scope
was chosen. The tool also allows switching between graphical and outline views at the click
of a button, which makes the generation of separate indented lists or tables unnecessary.
• Easy manipulation of objects (that is, editing of the hierarchy) is probably one of the
strongest arguments for MindManager. It is a simple matter to manipulate and update the
hierarchy by dragging any item or group of items (for example a whole Plan) to another
level. When the optional auto-numbering function is used, all levels are renumbered
automatically.
• Vertical or horizontal graphical views are available, with tasks shown in blocks, or simply as
text branches (the latter reduces visual noise and thus saves a lot of space).
• The HTA tree can be shown in outline view, with simple promotion and demotion controls,
with collapsing and expanding of levels. Tree branches can also be collapsed at any level for
overview.
• The analysis can be easily extended by adding Plan information to each level of sub-goals. A
simple notation (based on the recommendations in the literature) can also be added to
indicate Plan information. In addition, each sub-goal can be annotated with as much textual
information as needed. This can be extended to indicate very complex sequential and
conditional logic.
• MindManager can also support one of the most important aspects of task analysis: choosing
decomposition categories and developing task descriptions (Kirwan and Ainsworth, 1992).
The analyst simply adds the generic categories to the list of “map markers”. This list can
remain visible during the analysis to serve as a memory aid and the analyst can attach the
icon of any of these categories to the analysis or specific tasks (see Figure 2).
• Relationships between tasks can be indicated by means of a labeled line that links tasks (for
example, tasks that share the same resource).
• Static tables or dynamic spreadsheets can be inserted into tasks at any place in the hierarchy
to further extend the description of task attributes.
• Any number of task or resource attributes can be added to a task, for example roles,
environment or tools.
• By entering timing and resource information per task, a schedule (timeline) can be produced,
which can be exported to Microsoft Project.
• The tree can be filtered on a number of criteria, such as resource, label, etc.
• Reports can be produced directly from MindManager, or exported to Word or Excel for
further refinement.
• Macros and "Wizards" can be developed to guide the analyst in capturing of information and
to speed up the process of producing final reports.

Page 8 of 14
 Jacques V Hugo

• The whole or part of a hierarchy can be exported to Visio, HTML, Word, PowerPoint, Excel,
MS Project, bitmap or text outline.
• MindManager also makes provision for formal or informal review of analyses by allowing
reviewers to add comments to any part of the analysis, thus providing traceability.
• The HTA can be further enhanced and extended by indicating function allocation. This is
achieved by simply adding a marker (icon, text, etc.) for the allocated resource to the task.
This can be further enhanced, for example, by adding the basis for the allocation in the
descriptive text that accompanies every task.

4.3 Preparing to use MindManager for HTA

As a general-purpose tool, some initial work may be needed to set up a work environment in
MindManager that supports the unique requirements of HTA. Although optional, it is desirable to
ensure consistency. This usually involves creating a template with as many of the following
elements as needed (see Figure 2, Figure 3 and Figure 4):
• Decomposition categories
• Layout conventions (e.g. vertical or horizontal)
• Display conventions for task levels and Plans (font, color, numbering, etc.)
• Map markers (special icons and symbols, task categories, pre- and post-conditions, etc.)
• Resources (human and other resources)
• Macros and Wizards

However, it should be noted that in MindManager “local format” changes that are
independent of the template can be made on-the-fly at any time; the template simply provides
pre-defined styles, but does not restrict the representation of HTA information to those styles.

Page 9 of 14
 Jacques V Hugo

Symbols for generic decomposition categories can Some of the elements may be generic across all
be applied to any place in the HTA tree. These tasks; some may apply to specific operational
symbols are optionally attached to a sub-task on the scenarios, as shown in the following example:
tree and thus serve as mnemonics, which tell the
analyst that detail information is available for the
particular task:

Figure 2: Decomposition Categories

Special icons or symbols may be used to
graphically indicate task attributes such as goals,
sub- goals, task sequence (SEQ, OR, AND), pre-
and post-conditions, task triggers and terminal
tasks. As with decomposition flags, these symbols
are also mnemonics that provide information about
a particular task, without having to open another Figure 4: Task resources and conditions
window:

Figure 3: Task flow symbols

Page 10 of 14
 Jacques V Hugo

5. EXAMPLES OF USING MINDMANAGER AT PBMR

MindManager is used at PBMR throughout the task analysis process, but particularly in the
early stages when information is captured directly from the source documents and from the
subject matter experts, as described in par. 2.2.
The following examples have been extracted from actual analyses of PBMR tasks.
First, when the analyst focuses only on the highest levels of the hierarchy, a collapsed
diagram shows only goals and highest level plans and tasks:

Figure 5: Collapsed hierarchy

Next, the analyst can choose to focus only on a specific sub-goal, for example, “Control the
Helium Make-up System”. This diagram shows the entire sub-goal hierarchy, including
sequences and terminal tasks:

Figure 6: Plan, sequence and terminal tasks

Page 11 of 14
 Jacques V Hugo

The icons that appear on this diagram represent the task and decomposition categories shown
in Figure 2 and Figure 3. These are selected simply by clicking on the “Map markers” panel that
appears on the right hand side of the MindManager user interface. The same applies to the task
flow icons and the special task conditions shown in the example above.
When required, the entire hierarchy, or part of it, can be shown in outline format with task
flow symbols visible:

Figure 7: Outline format

The outline view can also be collapsed to any level:

Page 12 of 14
 Jacques V Hugo

Figure 8: Collapsed outline

Following is the complete graphical HTA for this task, shown to the third level:

Figure 9: Graphical HTA, showing 3 task levels

Page 13 of 14
 Jacques V Hugo

6. CONCLUSIONS

There is little argument among human factors practitioners about the validity and usefulness
of Hierarchical Task Analysis. However, to date there has not been a lot of consensus on the best
tools to use for the various analytical and representational phases of the process.
As shown in this paper, task analysis could be significantly easier with software support. But
as pointed out by Hone and Stanton, a tool would need very special features if it is to actually
help an analyst carry out an analysis. After all, no tool can replace the insight, skill and
experience of the expert analyst. It is true that "tools for depiction purpose only" offer very little
assistance to the analyst. Although generic tools can support the representation of task
information, complex task analysis in particular will benefit from the ability to input a dynamic
and changing structure. This is not provided by generic graphing tools. But even without the
features of specialized task analysis tools, the value of a tool that helps the analyst to capture,
organize and format task analysis information in a coherent, consistent, yet flexible manner, and
also eases the task of presenting information for reporting purposes, should not be
underestimated. If the tool can reduce the drudgery of these tasks and also supports the analyst’s
cognitive processes, it certainly deserves a second look.
It was demonstrated that a tool like MindManager goes far beyond a mere graphing tool like
Visio or TreeChart. Although it does not provide all the benefits of a specialized task analysis tool
like TaskArchitect, it provides real help to the analyst through its ability to make abstract
information visible, its ability to accommodate various layout styles and preferences, and to
manipulate the information in ways that support the production of a coherent analysis.
For human factors people who already own MindManager, it might be worth their while to
consider it as a tool that could be used to capture, analyze and document HTAs in an easy and
flexible manner, without sacrificing the rigour required for valid, coherent and consistent task
analysis. Even for those who do not own MindManager and are considering acquiring an
expensive, purpose-built software tool, a first look at MindManager should quickly demonstrate
the advantages.

7. REFERENCES

1. Ainsworth, L. & Marshall, E. Issues of quality and practicality in Task Analysis. Preliminary
results from two surveys. Ergonomics. 41, 11.1607 - 1617 (1998).
2. Diaper, D. & Stanton, N.A. The Handbook of Task Analysis for Human Computer Interaction.
Lawrence Erlbaum Associates, London. (2004).
3. Hollnagel, E. Task Analysis: Why, What and How. In: Salvendy, G. Handbook of Human
Factors and Ergonomics.3rd Ed. Wiley, Hoboken, New Jersey. (2006).
4. Lehto, M.R. & Buck, J.R. Introduction to Human Factors and Ergonomics for Engineers.
Lawrence Erlbaum Associates, New York. (2008).
5. Hone, G. & Stanton, N. HTA: The development and use of tools for Hierarchical Task
Analysis in the Armed Forces and elsewhere. Human Factors Integration Defense Technology
Centre. HFIDTC/WP.2.21/1 (2004).
6. Kirwan, B. & Ainsworth, L. A guide to task analysis. Taylor and Francis, London (1992).

Page 14 of 14

View publication stats