William Gerrie

1
William Gerrie Dal Vernon Reising

• Human-centered Alarm Sounds for Multi-Console
  Control Rooms

2
Presenters
William Gerrie
Control Systems Supervisor, ConocoPhillips
More then 25 years experience in Process
Automation (7 in RD/technology)
Project Manager on various projects including
automation lead
Ferndale Refinery www.conocophillips.com
3
Presenters
Dr. Dal Vernon Reising
Senior partner, Human Factors consulting group
based in North America
Specializes in human performance in process
industry operations
Former Principal Investigator of Honeywells
Abnormal Situation Management (ASM) Consortium
Human Centered Solutions Helping People Perform
www.applyHCS.com
4
Contributors / Co-authors

• Vicki Whittlesey
• Process Safety Education Director, ConocoPhillips
  Ferndale Refinery
• Kristi Hug
• Project Ops Liaison, ConocoPhillips Ferndale
  Refinery
• Peter Bullemer
• Senior Partner, Human Centered Solutions
• Chris Vazquez
• Project Ops Liaison, ConocoPhillips Ferndale
  Refinery

5
ConocoPhillips Ferndales Challenge

• Had a site-wide Experion PKS migration from
  legacy Honeywell systems planned
• Simultaneously relocating six individual
  unit-located consoles into a remote operations
  center, including centralized control room
• Variable practices in alarm sound use in the six
  individual unit-located control rooms
• e.g., the number of priorities audibly
  annunciated, the volume settings for the audible
  annunciations

6
ConocoPhillips Ferndales Challenge

• In 2006, in anticipation of this migration and
  move, Ferndales engineering and automation group
  collaborated with Human Centered Solutions
• As part of an ASM research project on effective
  methodologies for designing alarm sounds for
  multi-console control rooms

7
ConocoPhillips Ferndales Challenge

• In late 2007, coinciding with the control room
  relocations
• Ferndale deployed alarm sounds designed according
  to the methodologies evaluated, with the help of
  Human Centered Solutions (HCS)
• In early 2008, Ferndale and HCS evaluated the
  effectiveness of the alarm sounds
• Interviewed four of the six console positions
  relocated to the centralized control room
• Interviewed all four operating shift teams per
  position

8
Auditory Psychophysics

• Physical Dimensions
• Fundamental Frequency
• Amplitude
• Amplitude Envelope
• Spectrum Envelope

• Perceptual Dimensions
• Pitch
• Loudness
• Abruptness
• Timbre (quality)

• Pitch
• Higher frequencies perceived as higher pitch
• Loudness
• Higher amplitudes perceived as louder sounds

9
The Perception Human Factors of Sound

• 360 degree perception
• Unlike our visual attention, our auditory
  attention picks up information from all
  directions
• Primary frequency range
• We hear frequencies between 20-20,000 Hz
• Our ear is most sensitive between 200-1000 Hz
• Loudness
• Typical recommendations suggest a difference of
  15 dB between the background noise and signals
  that must be heard

10
The Perception Human Factors of Sound

• Critical Band Masking
• Lower frequency pure tones mask pure tones of
  higher frequencies
• Harmonics on the higher frequency tone offsets
  this effect

11
The Perception Human Factors of Sound

• Hearing Loss
• Whether from Aging or Industrial exposure,
  hearing loss tends to occur over the range of
  500-8,000 Hz
• This loss requires higher amplitude (loudness)
  for people to report hearing frequencies in this
  range
• Harmonics helps offset this effect, by helping
  people perceive the fundamental frequency in the
  range that they can hear

12
The Perception Human Factors of Sound

• Sound (and music) strongly influence our emotions
  and stress
• e.g., Musical scores in horror films
• Sound can strongly influence our mental workload
  and ability to concentrate
• e.g., driving, having a conversation with a
  passenger while listening to the radio, and the
  traffic starts to get heavier
• The onset / offset abruptness, compounded with
  loudness, can create startle effects
• e.g., square wave against perpendicular surface

13
Todays Typical Auditory DCS Alarms

• In multi-console control rooms, operators can
  have difficulty determining whose console is
  generating the alarm
• Standard alarms are abrupt, often loud
• Can generate a startle-reaction and
• Can contribute to heightened operator stress
  levels.
• Continuous, loud, high-pitched sounds can
  interrupt the operators ability
• To communicate with other people and
• To maintain their concentration
• Default Experion WAV file

14
Aviations Approach The Patterson Protocol

• The aviation industry has already developed a
  protocol for addressing many of these challenges
• Uses multiple acoustic properties
• To alert people to the urgency of the situation
• To attract attention without being aversive to
  the individuals responding an abnormal or
  emergency situation
• To easily identify the console with the alarm

15
Aviations Approach The Patterson Protocol
Pulse, Tone, or Note
Burst, Pattern, or Tune
Warning, or Alarm Sound

• Example Alarm Sounds via Patterson Protocol
• Priority Low High Urgent

16
HCS Approach for Multi-tone Alarm Sounds
Emergency Priority

• Each color represents a unique note (i.e.,
  fundamental frequency)
• Each fundamental is a frequency in our most
  sensitive hearing region
• Each note has multiple harmonics layered on top
  of the fundamental

Loudness
Time
High Priority
Loudness
Time
Low Priority
Loudness
Time
17
HCS Approach for Multi-tone Alarm Sounds
Emergency Priority

• Each note has an onset and offset time to
  eliminate abruptness startle
• For each console, the same 3-tone combination is
  used for all alarm priorities
• Priority is indicated by varying time durations
  of notes and pauses, as well as number of 3-tone
  combinations

Loudness
Time
High Priority
Loudness
Time
Low Priority
Loudness
Time
18
Operator Input Evaluation (1st Study)

• Ferndale and HCS conducted a study of parameter
  settings that most effectively communicated
  urgent, high, and low priority
• Length of tone itself
• Length of silence between tones
• Length of silence between bursts
• Operators from five console positions
  participated, spanning four shift teams
• The results were used to build unique alarm
  sounds for each of the six consoles

19
Operator Feedback Evaluation (2nd Study)

• Ferndale and HCS completed a follow-up evaluation
  after operators had been in the centralized
  control room (CCR) for at least several weeks
• Each console moved in at a different time, with
  the last console having several weeks of
  experience with the new alarm sounds
• Effectiveness issues evaluated included
• If there were issues with hearing the alarm
  sounds
• If sounds were confused with neighboring consoles
  
• If sounds were distracting from neighboring
  consoles
• All four shift teams for each of the four console
  positions in the CCR participated in the
  interviews

20
Alarm Sound Impact Ratings
21
Alarm Sound Impact Ratings Cont.

• Acceptability ratings by console positions
• Positive acceptability scores on most dimensions
  for all consoles
• All consoles had a positive, overall
  acceptability rating

22
Alarm Sound Impact Ratings Cont.

• Lowest scores related to alarm sounds
  discriminability
• Ability to recognize my alarm sounds as different
  from other consoles
• Other alarm sounds distracting during high alarm
  periods

23
Console Assignment Discriminability

• For initial Console assignment
• 3 of the 4 consoles had a common pattern between
  Notes 12 and 23
• This was contributing to the confusion, along
  with cross-room interference

A (4th)
B

• Studies from the medical field using the
  Patterson Protocol
• Suggest that people struggle with identifying
  individual notes
• We are better at recognizing different patterns
  of notes

C (3rd)
D (1st)

• In the subsequent Reassignment
• We have used more distinctive Note-to-Note
  patterns for greater disparity between consoles

E (2nd )
F
24
Summary of Operator Comments

• Most operators provided very positive comments
• Initially, operators were not aware of
  differences between consoles alarm sounds
• Did not really pay attention to it
• Not trained on sound characteristics
• After provided with an explanation, most
  operators realized that they had heard
  differences
• After reassignment of alarm sounds following 2nd
  evaluation, many operators indicated console
  discrimination had improved
• Ability to adjust volume was significant factor
  in ability to hear the alarm sounds at the
  console, while not interfering with communications

25
Summary of Operator Comments Cont.

• Inability to recognize alarm sounds as distinct
  from other consoles mostly occurred when at team
  table in center of room
• There was some confusion when similar console
  alarm sounds are coming from across the room, as
  opposed to left or right of console, as discussed
  above
• Better approach for off-console indication would
  be a light tower on each console
• Cannot turn up volume loud enough to be heard
  effectively throughout control room without being
  too loud at the console and for neighboring
  consoles
• At least one operator that self-reported hearing
  loss reported hearing the new sounds, whereas he
  claimed he could not hear the sounds in the unit
  control room

26
Alarm Configuration Findings

• Use of both audible active time setting and
  audible re-alarming important
• Alarms reconfigured for 15 second active period
• Audible re-alarming period set to 45 seconds
• Previous ASM study demonstrated that, on average,
  operators need 50 seconds to respond to an alarm
  this configuration is in line with that
• Each company and site would have to establish
  time frames that are suitable for the state of
  their own alarm rationalization practices

Silence

e.g., 15 sec
e.g., 45 sec
If still unacknowledged
27
Alarm Configuration Findings Cont.

• Make sure redundant servers have same ACK
  function settings
• Different settings at Ferndale caused audible
  alarm to annunciate one more time after
  acknowledgement
• Alarm sound should be set to 10 dB above ambient
  sound to ensure it can be heard above
  communications
• Provide operators with test sound to allow
  adjustment of volume (if adjustability is
  provided to them)

28
Conclusions

• ConocoPhillips and Human Centered Solutions have
  been successful in designing and deploying alarm
  sounds that
• Communicate more information
• Take advantage of Human Perception strengths,
  while avoiding Human Perception weaknesses
• Are well received by Operators
• In addition, Experion PKS users should consider
  using the Audible Active Time and Audible
  Re-alarm settings
• Make sure your alarm configuration settings are
  identical on servers for the same console