Alarm Systems

Abstract

In any control room alarm indications are installed to present particularly important information to the operator about process
deviations, plant disturbances and critical plant conditions. It is the operator’s responsibility to observe the patterns
of warning and alarm indications and to deduce the correct actions which are needed. Alarm systems have been a major concern
within complex industrial processes for many years. The Halden Project has developed many computer-based alarm system concepts
and performed a series of tests and experiments in HAMMLAB since 1983. This chapter provides a short summary of the different
alarm concepts and findings from tests and experiments. The experience that the Halden Project has gained in alarm system
development and testing have been instrumental in providing lessons learned and recommendations for alarm system design and
implementation for the industry.

Full text

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
ALARM SYSTEMS
Øivind Berg, Magnhild Kaarstad, Jan Erik Farbrot, Christer Nihlwing, Tommy Karlsson
OECD Halden Reactor Project
E-mail: oivind.berg@hrp.no, magnhild.kaarstad@hrp.no jan.erik.farbrot@hrp.no,
christer.nihlwing@hrp.no, tommy.karlsson@hrp.no
Belen Torralba
CIEMAT
E-mail : belen.torralba@ciemat.es
Abstract
In any control room alarm indications are installed to present particularly important
information to the operator about process deviations, plant disturbances and critical plant
conditions. It is the operator’s responsibility to observe the patterns of warning and alarm
indications and to deduce the correct actions which are needed.
Alarm systems have been a major concern within complex industrial processes for many years.
The Halden Project has developed many computer-based alarm system concepts and performed
a series of tests and experiments in HAMMLAB since 1982. This chapter provides a short
summary of the different alarm concepts and findings from tests and experiments. The
experience that the Halden Project has gained in alarm system development and testing have
been instrumental in providing lessons learned and recommendations for alarm system design
and implementation for the industry.
1. INTRODUCTION
Alarm systems have been of major concern within complex industrial processes for many
years. Within the nuclear community, the Three Miles Island (TMI) accident in 1979 was one
of the first major events that showed the importance of the human-machine aspects of the
systems in general, and the alarm system in particular. As the operators were trying to
understand what was happening in the plant, hundreds of alarms arrived. As a result offar the
accident it became evident that improvements had to be made in alarm processing and alarm
presentation. This was a big challenge to the nuclear community including the Halden Reactor
Project (HRP) which started a comprehensive research and development program on alarm
handling.
An alarm system should:
• Alert the operator to the fact that a system or process deviation exists;
• Inform the operator about the priority and the nature of the deviation;
• Guide the operator’s initial response to the deviation; and
• Confirm, in a timely manner, whether the operator’s response corrected the deviation.
This definition has served as a reference for many of the alarm-related projects performed in
Halden ([1], [2,] [3]).
Chapter 12 - - 1 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
The Halden Project has performed many experiments in HAMMLAB involving alarm systems
in the period from 1982 and up to now. These systems as well as experience from testing and
experiments are described in the following sections.
2. HANDLING ALARMS WITH LOGICS - HALO SYSTEM
The development of the Handling Alarms with LOgics (HALO) system was a major
undertaking by the Project in the beginning of the eighties. The main functions of HALO
system were to extract relevant alarms out of a large amount of process signals and to present
these alarms in a way that provided a clear overview for the operator[4]. As a result of this
alarm processing technique, a rather drastic reduction in the amount of alarms was obtained
compared to conventional alarm systems.
The HALO concept had three kinds of displays that the operator could request on different
screens: An overview display, detailed alarm group displays and alarm text displays.
The objective of the overview display was to give the operator a possibility to obtain with a
glance the main status as well as the alarm situation of the process. The overview was a
schematic diagram divided into areas representing sub-systems in the process. When one or
more alarms in a subsystem were active, the corresponding area in the overview display was
given the actual alarm colour, e.g. red for 1st priority alarms (See figure 1). The alarm group
detail displays were schematic diagrams which could display individual alarms in a way
similar to the overview. The alarm text displays were lists of alarm indications in
chronological order.
Figure 1: The HALO overview picture
The HALO system was used in a comprehensive experiment documented in [5]. This was an
experimental comparison of three computer based alarm systems using the NORS simulator.
The three alarm systems 1) NORS conventional alarm lists, 2) HALO text and 3) HALO
symbolic, utilized different degrees of filtering and graphic presentation.
Overall there was little observed difference in performance under the three display types. The
studies did not find that the HALO system offered the operator greater support than the
conventional system in detecting abnormal plant conditions. However, there was support for
Chapter 12 - - 2 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
the hypothesis that HALO helped the operator select the appropriate process display to
further analyse the disturbance.
In the diagnosis task, the operators using filtered alarms lists performed better than the
operators using an unfiltered alarm list [6].
In a later study of HALO [5], alarm information was integrated in the process overview and in
the process displays, reducing the number of levels in the display hierarchy from three to two.
The integration of alarm and process information was regarded much better than the earlier
separate presentation.
3. CRITICAL FUNCTION MONITORING SYSTEM - CFMS
A Critical Function Monitoring System (CFMS) is a good example of an alarm system for a
special purpose. This type of system has been regarded important for nuclear power plants.
The CFMS system provided information to the operator about the status of the plant in terms
of a limited set of indicators, termed critical functions. Status information about seven critical
functions and each critical function’s success paths – lower level indicators which, if within
established ranges ensure the integrity of the critical function – was available to the operators.
From a CFMS validation experiment performed in the Loviisa NPP in Finland [7], it was
concluded that the use of CFMS improved the condition of the safety functions.
The CFMS was most often used in connection with detection and confirmation of alarms, and
only infrequently used in the planning and decision for action. The operators used the process
diagrams for selecting the appropriate remedial operation.
The findings of the study point to a general issue with multi-function support systems: how to
integrate information which can be used for one purpose with the other purposes for which it
can, or is intended to be used. The operators demonstrated a good understanding of how to use
the information about critical functions in order to detect faults in the plant and to confirm
validity of the alarm information in itself. However, they did not integrate the information
from the CFMS directly with transient mitigation strategies, or used the system as a focal
point for planning and selecting which procedures to use. It seemed that the primary or direct
features of the CFMS displays were easy to understand, but that the secondary, indirect
features, in particular the success path, were more difficult. Even though this did not have a
strong influence on the rated usefulness, there appeared to be room for improvement of the
more complex features of the displays.
4. SUCCESS PATH MONITORING SYSTEM - SPMS
A Success Path Monitoring System (SPMS) is not an alarm system in itself, but it relates to
alarm systems and the kind of information one should present to operators.
The SPMS system implemented in HAMMLAB [7] provided on-line assessments in real time of
the status of success paths that are used to maintain critical plant functions. It presented
information about plant performance, in terms of success path system functioning and the
availability of systems that are used to achieve plant safety objectives. It also informed the
operators when entry conditions existed in the plant for utilising success path systems, as well
as conditions which required their termination, in order to prevent their use at times in which
damage could occur.
An experiment was performed [7] with one scenario and three experimental conditions: (1) the
HALO condition, (2) the CFMS condition and (3) the SPMS condition. In [7] the authors
Chapter 12 - - 3 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
concluded that the use of SPMS improved the performance of the SPMS group compared to the
two other groups”, as the SPMS subjects performed significantly better than both CFMS and
HALO objects. Overall the results clearly illustrated quite distinct advantages of the SPMS.
Speed and accuracy of operator performance in taking appropriate corrective action was
clearly superior with the SPMS and well up to prior expectations.
5. COMPUTERISED ALARM SYSTEM TOOLBOX - COAST
COmputerised Alarm System Toolbox (COAST) ([8], [9], [10]), is a generic alarm system
toolbox that enables configuration of intelligent alarm systems that can be adapted to the
specific process at hand. COAST can be used as a stand-alone tool, or it can be integrated into
a conventional process control system.
COAST consists of an off-line alarm system definition part, an on-line alarm system processing
part, and an alarm extraction part. The system utilises measurement signals and binary
signals from the process control system to generate alarms. COAST then filters and
suppresses the alarms based on simple or advanced algorithms, and it generates new
aggregated alarms for better explanation to the operator. It will send the updated alarm
information to the control system for presentation on the operator’s interface. COAST also
structures these alarms into different types of alarm lists. Examples of aggregated alarms are
function-oriented alarms like safety function alarms, model-based alarms for early fault
detection, or high-level alarms describing the process state, which in turn are used for
suppressing other non-relevant alarms. The COAST features and functionality is further
described in [10]. The report also discusses and explains how COAST fulfils requirements and
recommendations specified for enhancing alarm systems.
Several Halden Project member organisations and companies outside the Halden Project are
currently utilising COAST, either for making alarm systems, or for making other kinds of
event detection systems. In addition, all HAMMLAB simulators use COAST for alarm system
implementation and execution. COAST has shown to be very valuable for practical alarm
system projects and its flexibility seems to be appreciated. The HAMMLAB alarm systems,
which are continuously being improved, have benefited from the flexibility offered with respect
to modifications and maintenance as well as functionality. The development of COAST has
been driven by input from the COAST users, as well as established requirements and
recommendations to alarm systems in general.
6. COMPUTERISZED ALARM SYSTEM FOR HAMMLAB - CASH
Computerized Alarm System for HAMMLAB (CASH), an advanced alarm system utilizing
different alarm processing and presentation techniques, was the first application using
COAST.
The scope of CASH was two-fold: To provide good alarm processing and presentation using the
best combination of existing and new features, and on the other hand, allow experiments on
alarm systems, offering a broad range of different alarm processing and presentation
techniques [11].
The first prototype of CASH was developed in 1992-1994, and an interface with two
hierarchical levels of information was made. Level 1 was the overview display that supplied
the operators with plant wide key process information and non-suppressed alarms. All
irrelevant information was removed from the overview level to avoid information overload.
Level 2 is composed by alarm selective displays, which show more detailed alarm lists (see
Chapter 12 - - 4 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
Figure 2). Two screens were used for the overview, one for the primary and one for the
secondary side. In addition, alarms were also presented in process displays.
Figure 2. Left: Level 1, Overview display, Right: Level 2, Alarm list display (primary side)
The goals of CASH with respect to alarm reduction were to devise methods, so that the set of
defined alarms are only presented when they really are alarms according to the alarm
definition. Also, it was important to be able to test varying suppression levels, and to group
and sort alarms in such a way that the presentation is optimized with respect to the operators’
information needs.
Structuring provides a method to reduce the information load on the operators without
removing any significant alarm information from the alarm system itself, and several alarm
reduction techniques are provided
CASH was implemented in HAMMLAB in 1995, and used for the first time in “A study of
control room staffing levels for advanced reactors” [12] for USNRC in 1995. In the fall of 1996
a large alarm experiment was executed, addressing alarm presentation, alarm suppression
and alarm availability. The experiment was a joint project between USNRC and HRP [13].
The alarm experiment compared different alarm prioritisations, suppression and display
types. Through interviews and debrief with the operators, valuable comments regarding their
preferences with regard to alarm presentation and profcessing were expressed. However, the
results showed that the operator performance was not markedly different for any of the
conditions. One of the explanations for these inconclusive results was that in this control room
setting, the highly skilled professional operators who participated in the study might have
been able to compensate for any differences in alarm systems design by using alternative
information sources, such as process displays and trend graphs.
7. HAMMLAB BOILING WATER REACTOR ALARM SYSTEM - HAMBO
The development of an advanced, fully computerised alarm system for the HAMmlab BOiling
Water Reactor simulator (HAMBO) was ordered by the Swedish and Finnish BWR utilities
and developed in co-operation with IFE. The project started in January 2001 and was finalised
in August 2002.
The purpose was to design a flexible alarm system possessing capabilities for performing
experiments with different solutions regarding alarm processing and presentation.
Chapter 12 - - 5 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
Alarms are presented in many ways:
• alarm lists
• alarm tiles
• alarms integrated in process displays
• alarms integrated in overview displays
• alarms integrated in other types of information systems
Software systems used for the development were COAST, COPMA, Picasso-3 and the
Integration Platform (IP), all developed by the Halden Project [14].
Each alarm signal in HAMBO is assigned a priority,and each priority is assigned a colour code
were first priority alarms are red while second priority alarms are yellow and third priority
alarms are green.
Below is a picture of the main alarm list and a process display including alarms presented as a
frame around the object/ measurement that is alarmed, coloured in the alarm priority colour.
In each process display, a small list of all active alarms in that display can be presented in the
bottom of the display at operator request.
Figure 3: Left: HAMBO alarm list, right: HAMBO process display with alarms
The large screen indicates that alarms exist in different system parts, by presenting the
alarm/ alarms with a frame (red, yellow or green) around the affected system. The triggering
condition for scram is also shown at dedicated areas on the large screen.
Figure 4: HAMBO large screen display
Chapter 12 - - 6 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
The HAMBO alarm system has also an alarm navigation display, indicating with a dot
coloured in alarm colour at the display/ system button that there is an alarm in that actual
display/ system. For further details on the design of the alarm system, please refer [14].
Having an electronic alarm system gives the possibility to build some logic into the system. In
the HAMBO alarm system, suppression logic is built into the release of a reactor trip, thereby
making e.g. the alarm list more usable for the operators. Also, electronic alarm response
procedures are developed in the HAMBO alarm system.
The HAMBO alarm system, has gone through a Human Factors evaluation [15], as well as
usability test [16].
The main conclusion from the review and the usability test of the alarm system was that the
HAMBO alarm system must be appreciated as an acceptable system, as it contain most
elements considered important for useful, usable and acceptable alarm systems.
The feature that was especially useful was the overview display and the alarm navigation
display. In the alarm navigation display, the operators can address system buttons where
alarms are indicated, and have a direct access to a list of alarms that are active in that
particular system. In this display it is also possible to acknowledge alarms system by system,
which provide the operators with a better overview. Also the electronic alarm response
procedures were useful for the operators. It saves a lot of time to address an alarm and read
the procedure on the screen than to find the right folder and the right page where the alarm
response procedure is described.
Both the usability evaluation and the human factors review concluded in some
recommendations for the further development of the HAMBO alarm system. The most
important point is that there are too many alarms in the system, and that the prioritisation
and presentation of alarms could be improved.
Another lesson learned from the evaluation, was that it is still a challenge to find ways to
present alarm information in computer-based control rooms in such a way that the important
alarms are not missed by the operators.
8. AN MFM BASED ALARM SYSTEM
A continuation of the HAMBO alarm system development was an integration and test of a
more advanced alarm system in cooperation with Swedish and Finnish utilities and a private
company, GoalArt. The alarm system, later called the GoalArt based alarm system, was
developed using GoalArt’s tools, a Multi level Flow Model (MFM) and State Based Alarm
Prioritization (SBAP). The main objective with the GoalArt based alarm system is to support
operators in their analysis tasks by separating causes from effects. The system identifies root
cause (primary) alarms and consequence (secondary) alarms, and provides information about
which secondary alarms are consequences of a specific primary alarm [17]. In addition, the
GoalArt system identifies process components with a state different from what the system
believes to be the correct component state, and indicates what it believes to be the correct
state. The GoalArt based alarm system is also able to mark objects as “out-of-operation”, for
instance because of ongoing maintenance work, and no alarms related to that object will be
generated.
Each alarm from the GoalArt system is further assigned a priority. The priority is dynamic,
and may change as the disturbance develops. Four priorities are used. Priority 1 identifies the
most critical alarms, and priority 2 and 3 identifies alarms with lower criticality. Priority 4
Chapter 12 - - 7 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
identifies the lowest criticality which is classified as events and is not included in the alarm
list or the process formats.
When building the model for implementation in the HAMBO simulator, safety action chains as
turbine trip and scram were modelled to begin with.
Figure 5: Left: GoalArt alarm presentation in a process display. Right: GoalArt alarm list
In the process displays, the primary alarm is presented as a triangle with an explanation
mark inside, while the secondary alarm is a red frame triangle (Figure 5). In the GoalArt
alarm list there are three fields. In the “Critical alarms” field on the top, all incoming alarms
related to safety actions arrive. The second field is for “Primary alarms” and the last for
“Secondary alarms” (Figure 5).
The GoalArt alarm system was evaluated in two phases; an explorative test and a final
usability test [17]. The inputs given by operators in the explorative test were taken into
account before the final usability test was conducted. From the final usability test, several
observations were made. First of all, the GoalArt based alarm system gives the user a unique
support in observing, interpreting and diagnosing a disturbance.
The operators stated that if this system had worked as intended, it would have been a really
good alarm system, as it is designed to identify the causes, not the consequences. The GoalArt
alarm system solves many of the current problems with alarm systems today. It does not
overload the operators with alarms that are only events or consequences, and the way alarms
are presented, in a primary list and a secondary list, makes it easy to detect new events that
arrive, and to take corrective actions in time. The operators expressed that they felt that they
are in control when using this alarm system.
The most important identified needs for improvement with this alarm system were that more
plant states need to be defined, as the Goal Art alarm system is not working well in situations
between defined plant states. In addition, the model needs to be tuned better, as it sometimes
give a wrong alarm. Logging of alarms should also be included, as it is essential to be able to
go back and analyse events when something has happened in a plant.
9. USE OF HAMMLAB DATA FOR ALARM ANALYSES1
The role and the use of alarm systems have been studied by means of an experimental
analysis of operator visual activity, using eye movement tracking equipment [18]. A precedent
1 This section is by Belén Torralba, CIEMAT, Spain
Chapter 12 - - 8 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
work on exploring the role of alarm system was carried out by OECD Halden Reactor Project
[19]. The eye movement tracking equipment consisted of two ASL model 4000 SU eye-
movement trackers, providing real-time measurement of the point of gaze at 50Hz, a PC with
monitor, two video monitors and a calibration surface. A head-mounted scene camera captured
the situation viewed by the participant.
The data were collected in the full scope Nokia Research Simulator (NORS) PWR simulator of
the Halden Man-Machine Laboratory (HAMMLAB), during an explorative experiment that
addressed how automation influences operator performance in nuclear power plants control
room.
Two main variables, which are based on eye movement tracking data, have been measured:
the number of visual accesses per time unit (frequency data), and the percentage of time
gazing at areas of interest (duration data). Relevant data about the main alarm system
features are considered such as the number of active alarms, the number of alarms existing at
the beginning and at the end of each period, the operator interactions with the alarm system
(silence and acknowledgment actions), etc.
The main results of three scenarios indicate that the mean of the percentage of time gazing at
the alarm system was approximately 10% of the total available scenario time. The alarm
system was first of all used in the detection phase while the operators spent major time to
gather information using other information systems for fault diagnosis and implementation of
operator actions. There are high correlations between the number of visual accesses to alarms
displays and the percentage of time gazing at alarm displays. There is a general tendency that
reactor operators (main responsible for reactor safety) spend more time and present a higher
number of visual accesses looking at the alarm displays than turbine operators.
The expected output of the project will be on contributing with empirical data on the role and
the use of alarm system, to give insights into the human performance patterns on complex
situations, and to provide design recommendations for alarm system interface.
We would like to point out the feasibility of using data previously collected at HAMMLAB for
re-analyses of retrospective data with a different purpose than the scope of the experiment. We
would also like to express our acknowledgment to the Halden Project staff.
10. SUMMARY
The Halden Project has experience from design and implementation of various types of alarm
systems in HAMMLAB. In addition, the Halden Project has gathered substantial data through
experimental studies of operator performance and usability evaluations of various alarm
systems. An advantage with experimental alarm systems is that reconfiguration and testing of
different presentation techniques can be made without any risk to an actual plant, and
without too much effort.
Alarm systems do not always work as intended during disturbances. The main reason for this
is that they provide too many alarms to the operator, leading to cognitive overload. In parallel
to development of alarm systems, over the past few years therefore several sets of guidelines
or recommendations for alarm systems have been developed. While the early recommendations
concentrated on alarm suppression requirements and display methods, the recent ones in
addition stress purpose and lifecycle management of the alarm system. Performance
monitoring of alarm rates is one aspect of management of the system. System management
issues are important to ensure that maintenance gives continuous lifecycle improvement
Chapter 12 - - 9 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
rather than degradation. A better alarm system may result in more stable operation of the
process, and better economy for the plant.
The experience and knowledge obtained from development and evaluation of various alarm
systems in HAMMLAB have been used for many purposes. Licensing authorities use the
experimental findings as basis for writing guidelines. The utilities take advantage of specific
alarm design proposals in HAMMLAB when modernising their plants. IFE has also supported
many non-nuclear industries, especially the petroleum sector to improve their alarm systems.
Although various guidelines with regard to design, generation, structuring, presentation, and
implementation and management of alarm systems exist, there is still a need for further
improvement and innovations to design useful alarm systems for the operators. A summary of
“Recommendation to alarm systems and lessons learned on alarm system implementation” is
given in [20].
11. REFERENCES
[1] NUREG-0700, Rev 1, (1996). U.S. Nuclear Regulatory Commission: Human–System
Interface Design Review Guideline.
[2] O’Hara, J.M., Brown, W.S., Higgins, J.C., Stubler, W.F.: Human Factors Engineering
Guidance for the Review of Advanced alarm Systems. NUREG/CR-6105. Brookhaven
Nagional Laboratory, 1994.
[3] W.L. Rankin, E.G. Duvernoy, K.R. Arnes, M.H. Morgenstern, R.J. Eckenrode:
Near-Term Improvements for Nuclear-Power-Plant Control-Room Annunciator Systems
NUREG-3217, 1983.
[4] Visuri, P. Øwre F. (1982). Forming and presenting process control room alarms using
computers – The HALO concept. HPR-283.
[5] Baker, S., Gertman, D., Hollnagel, E., Holmström, C., Marshall, E., Øwre, F. (1985). An
experimental comparison of three computer-based alarm systems: Results and
conclusion. HWR-142.
[6] Hollnagel, E. Øwre F. (1984). The NORS/HALO system: Background and methodology for
experiment 1. HWR-90.
[7] Baker, S. Marshall, E. Reiersen, C. Smith, L. Gaudio P.J. Jr. (1988). The Experimental
Evaluation of the Success Path Monitoring System – Results and Conclusions. HWR-
224.
[8] Bye, A. Handelsby, F. Storberget, T.W. Nilsen, S. (1994). COAST – Alarm system toolbox
for the process industry. HWR-389.
[9] Bye, A. Farbrot, J.E., Lee, D-Y. Louapre, G. (1998). COAST – Status and new features of
the alarm system toolbox. HWR-512.
[10] Bye, A. Farbrot, J.E., Nystad, B.H. Jakobsen, R. (1999). Problem areas with alarms
systems in the process industry – solutions utilising the COAST alarm system toolbox.
HWR-580.
[11] Moum, B.R., Førdestrømmen, N.T. Saarni, R. (1998). Alarm system CASH: Lessons
learned based on operators feedback and designers’ experiences. HWR-535.
Chapter 12 - - 10 -

The History of HAMMLAB - 25-Years of Simulator-Based Studies. Draft
[12] Hallbert, B. P. Sebok, A. Moriseau D. (2000). A study of control room staffing levels for
advanced reactors. NUREG/IA-0137.
[13] O’Hara, J. M. Brown, W. S. Hallbert, B. Skraaning, G. Persensky, J. J., Wachtel, J. (2000).
The effects of Alarm Display, Processing, and Availability on Crew Performance.
NUREG/CR-6691
BNL-NUREG-52600 US Nuclear Regulatory Commission, Washington, D.C.
[14] Karlsson, T., Meyer, B.D., Jokstad, H., Farbrot, J.E., Hulsund, J.E. (2002). The alarm
system for the HAMBO BWR simulator. HWR – 702.
[15] Kaarstad, M., Seim, L.Å. (2004). Human Factor evaluation of the HAMBO alarm system.
IFE/HR/F-2004/1224
[16] Kaarstad, M. (2004). Usability evaluation of the HAMBO alarm system. IFE/HR/E-
2004/028
[17] Kaarstad, M., Nihlwing, C. (2007). Usability evaluation of the GoalArt based alarm
system. IFE/HR/F-2007/1318
[18] Torralba, B., Martínez Arias, R. and Solá, R. (2007). An Experimental Study of Operator
Activity using an Advanced Alarm System in Nuclear Power Plant Control Room en
Boring, R., Hallbert, B., Persensky, J.J., Blackman, H., Flores, T. (eds) Joint 8th IEEE
Conference on Human Factors and Power Plants and 13th Annual Workshop on Human
Performance / Root Cause/ Trending / Operating Experience / Self Assessment,
Monterey, CA, USA. 25 – 31 August.
[19] Skraaning jr, G. Andresen, G. (1999). The function of the alarm system in advanced
control rooms: An analysis of operator visual activity during a simulated nuclear power
plant disturbance. Paper at EHPG, Loen, 1999.
[20] Sørensen, A., Veland, Ø., Farbrot, J.E., Kaarstad, M., Seim, L.Å., Førdestrømmen, N.,
Bye, A. (2002). Recommendations to alarm sytems and lessons learned on alarm system
implementation, HPR-354.
Chapter 12 - - 11 -