A New Method for Alarm Rationalization

1
A New Method for Alarm Rationalization

• 2nd Annual ISA Safety Division Symposium
• May 2006, Houston, Texas

2
Presenter

• Bill Hollifield
• Principal Alarm Management Consultant, PAS

www.pas.com
3
Alarm Rationalization

• The DCS Alarm Problem In A Nutshell

Thousands of Alarm Events Cannot be Evaluated By
The Operator! A Highly Hazardous Situation
4
Alarm Rationalization

• Alarm Rationalization A Rigorous, Powerful, Best
  Practice Methodology That Achieves Excellent
  Results When Done Properly
• Quotes from operators after alarm system
  improvement projects
• Finally the alarm system makes sense.
• The alarm system is useful now. It sure wasnt
  before.
• You can understand the alarms now they have
  real meaning.
• This is the best thing we have ever done.
• Im not constantly dealing with a bunch of
  incomprehensible
• alarms anymore.
• The alarm system is now under control!

5
Conventional Alarm Rationalization What Does It
Take?
Engineer Jane We just finished making a big
improvement in our alarm system. We did something
called Alarm Rationalization.
Engineer Jim Whats involved? (lengthy
conversation ensues)
Well, that sounds interesting. Well have to
look into that. And that reminds me, I really
need to schedule my colonoscopy.
6
Alarm Rationalization You Know You Need To Do
It!

• All the Experts Say Its Needed
• Avoid Very Serious Problems in the Future
• Ensure alarms are engineered properly
• Ensure consistency in alarm settings
• Eliminate duplicate alarms
• Ensure proper and meaningful Priority and Trip
  Point settings
• Create a Master Alarm Database, a reference for
• Real time alarm management
• Audit / Enforce and Management of Change
• Dynamic State-Based Alarm Management
• Flood suppression
• Operator information

7
Alarm RationalizationWork Process (Highly
Condensed)

• Evaluate Every Possible Alarm To Insure That
• All Alarms Require Operator Action
• Alarms are based on the best indicator of the
  root cause of an abnormal situation (and not
  duplicated)
• All Alarms must be produced upon abnormal
  situations only, not from normal situations
• Alarm Settings (not just trip points) are
  properly designed
• Alarm Priorities are based on a sound and
  consistent methodology

8
Alarm RationalizationPriority Determination

• Typical Grid-Based Priority Determination

Severity of Consequence, Plus
Determines Alarm Priority
9
Alarm Rationalization Resources
AlarmAnalysis
PIDs
PointConfigurationData
Process History
Plant Experience Knowledge Process, Equipment,
Operations, Procedures
Documents PHA/SOP/EOP/PSM/LOPA/SIL/HAZOP/

• Board Operators
• Process Control Engineers
• Safety, Health, Environmental
• Production Maintenance Engineers

RationalizationSoftware
10
Alarm RationalizationIf Its So Great, Why the
Resistance?

• Less than one in 10 facilities that acquire alarm
  analysis software proceed with Alarm
  Rationalization. Why?
• Alarm Rationalization is Expensive and Disruptive
  to Plant Operations
• Four to six people required, who are
  knowledgeable in the process being examined
• Such people are in short supply (downsizing)
• Typically weeks of concentrated effort per system
  
• Other work falls behind
• Process is tedious and unpleasant

11
Alarm RationalizationHow Long Does It Take?

• Typical System 4000 alarmable tags
• Overall rate 100 Tags per full day of
  Rationalization Meetings
• Rate depends on finding substantial equipment
  duplications (50)
• 40 Working Days (8 weeks) of Rationalization
  Meetings will delay other projects
• Significant additional time must be spent outside
  of meetings

With 4 to 5 people, internal cost can exceed
100,000 in direct cost, plus opportunity cost!
12
Alarm RationalizationRequirements for an
Improved Method

• No sacrifice in safety
• No sacrifice in performance improvement of the
  post-rationalized system compared to the
  standard methodology
• Significantly increase overall productivity and
  reduce cost
• Quality of the process must embody superior
  engineering principles. A change from the
  standard process must be backed up by data

How can this be done???
13
Focused Alarm RationalizationThe Data Speaks
Loudly

• Since the publication of EEMUA 191, terabytes of
  real-world alarm system performance data have
  been gathered by PAS and evaluated.
• A surprising finding Given long-period data of a
  typical system (a year or more), only about 25
  to 40 of configured alarms produce even a single
  alarm event.
• Therefore, 60 - 75 of the time spent in
  Rationalization is discussing alarms that have
  zero impact on the alarm problem.
• Only about 10 of the alarmable points produce
  more than 10 individual alarm events - and thus
  up to 90 of the points have almost no impact on
  the system performance at all.
• With several safeguards and refinements, this is
  the core concept of Focused Alarm
  Rationalization.
• Focus on solving the alarm problem

14
Focused Alarm RationalizationAlarm Frequency
Distribution
Most configured alarms produce very few or no
alarm events.
Scale is reduced, events peak at over 28,000
Only 365 of the 2,531 Alarms Produce more than
10 alarm events per month
1,576 of the 2,531 Alarms Produce less than one
alarm per month!
2000
2,531 Configured Alarms (39) of 6,486 Alarmable
Points, Producing 1 or More Events. 61 produced
zero events.
15
Focused Alarm RationalizationTypical Alarm
Frequency Distribution
Full Alarm Rationalization
7000
6000
100
5000
Focused RationalizationNo Threshold
4000
FocusedRationalizationWith Thresholds
3000
39
2000
14
1000
6
0
PointsProducing
PointsProducing
AlarmablePoints
PointsProducing
At Least 1 Alarm
More Than 1
More Than 10
Alarm Event
Alarm Events
Event in a Year
per Month
Per Month
16
Focused Alarm RationalizationA 5-Step
Methodology

• Step 1 Adopt An Alarm Philosophy
• Adopt a comprehensive Alarm Philosophy, with
  ongoing analysis requirements, rationalization
  method, configuration rules, etc.
• Step 2 Determine the Focused Alarm
  Rationalization List
• Limit Alarm Rationalization to the points that
  have produced any alarm events in an appropriate
  analysis period. (Most Conservative Option, no
  Threshold)
• A comprehensive alarm analysis spanning 6 or more
  months is used to produce this listing a year is
  far preferable. This allows for any seasonal
  effects on the alarm lists to be found.
• The resulting list will typically be about 25 to
  30 of the total alarmable points.

17
Focused Alarm RationalizationA 5-Step
Methodology

• Step 3 Add To the Focused Alarm Rationalization
  List
• ALL currently configured Emergency (highest DCS
  Priority) alarms that were not included in the
  above count.
• All required alarms from PSM, PHA, SIL, LOPA,
  etc.
• These additions will likely add less than 5
  percentage points to the Focused DR point count.
• Step 4 Additional Safety Review and List
  Additions
• The actual alarm configuration found on the
  system in Step 3 may not accurately reflect what
  has been specified to be there, due to improper
  or lacking Management of Change processes.
• Therefore, conduct a specific, focused review,
  with unit-knowledgeable engineers and safety
  specialists. Identify any alarms that are not
  currently configured as Emergency priority, that
  should be. Reference recent Process Hazard
  Analysis or similar safety studies. Point count
  addition will be negligible.

18
Focused Alarm RationalizationA 5-Step
Methodology

• Step 4 (Continued) Perform Alarm Rationalization
  on the list
• At this point the total list of tags to be
  rationalized is likely only 30 to 40 of the
  conventional method list.
• Perform Alarm Rationalization on this list in the
  conventional manner.
• Implement the changes on the DCS.
• All pre-determined alarm settings in the Alarm
  Philosophy will be implemented as well, but this
  does not require time in the Rationalization
  meetings.

19
Focused Alarm RationalizationA 5-Step
Methodology

• Step 5 Ongoing Monitoring
• Initiate an on-going monthly program to check the
  alarms produced in the prior month, and
  rationalize any new ones that occurred that
  were not included in the steps above.
• Alarm analysis software makes identification of
  these points a simple task, taking only minutes.
• Then, short rationalization meetings of an hour
  or two can handle the small number of new
  points involved, without significantly disrupting
  work schedules.

20
Focused Alarm RationalizationExpected
Performance Results

• After Focused Rationalization, every single alarm
  that has occurred for many months has been
  rationalized.
• The point count, time, effort, and cost is
  reduced more than 50 (usually 65 to 70 or
  more) compared to the conventional method.
• All tags that are part of the alarm problem
  have been addressed - the system dynamic
  performance is improved by the same amount as in
  Conventional Rationalization
• With the same effort as a Conventional
  Rationalization for one system, three or more
  systems could be substantially improved using
  the Focused Rationalization method.
• New alarms are rationalized the following month
  after they occur, an acceptable tradeoff.
• 100 of all Emergency priority alarms, and other
  required alarms, have been rationalized, and
  special effort made to insure that the total list
  of those is proper.

21
Focused Alarm RationalizationScope Savings A
Typical Example
Total Alarmable Tags 6,486
Unique Alarm Events over 270 days 463,536
Average Alarms Per Day 1,717
Unique Tags Involved in all these events 2,170
Percentage of Tags producing even 1 alarm event compared to all Alarmable Tags, plus all E Priority, PSM, PHA, SIL, LOPA, etc. 35
Comparison Conventional Focused
Tags to Rationalize 6486 2270
Optimistic Duration (Days) 65 23
Internal Resource Cost at 1800 / day for meetings 117,000 41,400
Savings 42 Days and gt75,000
22
Focused Alarm RationalizationConfirm Before You
Start
Analyze your own data You can easily and quickly
verify whether the Focused Rationalization
methodology applies to your own system, before
you ever begin Rationalization.
23
Focused Alarm RationalizationDisadvantages

• Alarms not subjected to the Focused
  Rationalization process are initially left
  unchanged. New alarms do not have a
  rationalized priority, and may be inconsistent
  with the priority philosophy incorporated in the
  rationalized system.
• However, the worst case difference is only that
  an alarm might come in as a Low priority instead
  of a High, since all Emergency have been
  addressed. This gets corrected in the following
  month, Step 5.
• In conventional DR, tags without configured
  alarms are occasionally found that do need
  alarms. This methodology may not find them.
  However, Step 4 of the 5-point process does help
  address this and they would not be found if there
  were no Rationalization at all the real-world
  choice.

24
Focused Alarm RationalizationEven Larger
Savings are Possible

• Concept Introduce a Threshold. Even a small
  threshold will reduce the rationalization list
  substantially.
• Example Eliminate alarms that actuate less than
  once per month clearly not substantial
  contributors to the alarm rate problem.

Comparison Conventional Focused (ALL Alarm Events) Focused (gt1/month Frequency)
Tags to Rationalize 6486 2270 885
Optimistic Duration (Days) 65 23 9
Internal Resource Cost at 1800 / day 117,000 41,400 16,200
Savings 0 gt75,000 gt100,000
Using this method, 8 systems could be
substantially improved for the conventional price
of one.
25
Focused Alarm RationalizationUsage Requirements

• A Patent is pending relative to this methodology.
• PAS desires the maximum improvement of as many
  alarms systems throughout the world as is
  possible.
• Therefore, PAS will provide a license for the use
  of this cost-saving methodology free-of-charge
  for DCS-owning companies, in exchange for
  information on how much time, effort, and money
  is saved for each system that is rationalized in
  this manner. Contact us for further information.
• Non-DCS owners, contractors, service companies,
  and practitioners desiring to use the technique
  should contact us for licensing information.

26
Focused Alarm RationalizationConclusion

• The goal of a company with an alarm problem is to
  solve the problem.
• In Focused DR, you are always discussing points
  that contribute to your alarm problem.
• In a conventional DR you might be spending up to
  90 of the very expensive meeting time discussing
  points that have no effect on your alarm problem.
  
• Is documenting them a good idea? Of course. Is it
  worth the cost and effort? You decide.
• An alarm system is far, far better off after
  Focused Rationalization than after no
  Rationalization, which may often be the realistic
  choice.

27
Focused Alarm Rationalization

• Questions?

Available atwww.Amazon.com