Transcript Reliability from Data (English)

Reliability from
DATA
A framework for technology
OMDEC
1.
2.
3.
4.
5.
Maintenance / Asset Management Consulting
Training Programs
Software Tools
“Living RCM”
Canadian Company: Ottawa, Montreal, Toronto,
and Australia Locations
Sample Industries: Mining, Oil & Gas, Utilities,
Fleets, Government and Military
Why collect data?


Only one reason: To perform analysis. “Reliability Analysis”
Why analyze?
 To
improve the process of maintenance continuously.
(CPI = Continuous Process Improvement)

Why CPI?
 That’s
our (i.e. everyone’s, particularly
management’s) job.

Why?
 Economic
survival of the fittest. Keep up with change.
The “false” promise of CBM technology

Based on the logic that:
 The
more data the better,
 The faster the better, and
 The more views (PDAs, iPhone, etc) the better.
 All of the above are good, but there is a flaw in the
logic.

What is the logical flaw?
 There
is an infinite supply of the wrong data.
 The logic skirts the question: “What is the right data?”
What’s the right data?

Age (“life”, “life cycle”, “event”) data

Failure Mode occurrences with attributes:



event type (PF, FF, S, …),
RCM reference,
working age
Work orders

Condition monitoring data


relevant to the failure modes of interest.
RCM knowledge of failure modes.
RCM
Achieving reliability from data
Four challenges must be overcome:
1.
2.
3.
4.
Data extraction and transformation
Management of the work order –
RCM relationship
Sample generation
Typical focus
Reliability analysis
Unified EXAKT
Process
•Systematic
•Quick
•Results oriented
Challenge 1 Data extraction, transformation
Input from
CMMS
Data
transformations
Example: FMEA extraction
Output for
LRCM
Input from RCM Cost,
RCMO, RCM Toolkit, etc
Example: Work order
extraction
Input
from
Ellipse
CMMS
input
Data
transformations
Challenge 2 LRCM
…
the most difficult of the four - the key challenge
Text of the selected
knowledge record
KPIs
Event type indicators:
PF (blue), FF (red), S
(yellow).
Add/Edit KRs (with
audit trail)
“Slice and dice”
Text of the selected
work order
Dynamically,
1. Link the work orders and knowledge base.
in the day-to-day work order
2. Build the knowledge base…
process
Challenge 3: Sample generation
RCM Knoweldge
base
Work Orders that
have been linked
to the KB
Events table
(the sample)
Sample generation
/Challenge
3 cont’d:
CMMS Work orders
Events table
EF15
Work ord. 1, FF RCMREF15
EF16
Work ord. 2, FF RCMREF16
B16
EF16
Work ord. 3, FF RCMREF16
B16
ES15
Work ord. 4, S RCMREF15
B15
EF15
Work ord. 5, PF RCMREF15
B15
Legend:
Life cycles:
Left Suspensions:
Right (Temporary) Suspensions:
EF: endings by failure
ES: endings by suspension
Sample
Calendar Time
B15
Hazard model
0.781 t 
h(t ) 


2709  2709
0.7811
e0.06944MaxWSDrop 
Challenge 4:
Reliability
analysis and
EXAKT
+
Predictive Model
Predictive
model
RULE and
Confidence interval
+
Cost model
Decision based on:
Probability
EXAKT Decision based on:
Scatter
Cost and
Probability
RULE
Challenge 4 - Achieving
Reliability from data in EXAKT
Age data (CMMS)
CBM data
Cost data
Hazard model
Transition model
Cost, Availability,
Profitability model
RULE
Maintenance
Decision
Supplied by user
Modeling Software
Intermediate results
Final Result
Challenge 4 - CBM+Simulation in SPAR-PHM
No
maintenance
Replace radio
now
And plan overhaul
in 6 months
Projected worst
actor following
overhaul
OMDEC methodology
“living reliability”
“on-the-job”
Iterative
Integrated
LRCM Pilot On-the-job process
Overcoming Key Challenge 2
Team
1.
2.
3.
4.
5.
6.
Monitor work orders & KR links
Monitor knowledge record updates
Ask questions
Propose changes
Get feedback
Get consensus.
On the job teamwork
Progress reports
KPIs
LRCM
OMDEC
guidance
LRCM specialists
+
Company’s
Work orders
Methods,
and KR links Engineers, planners,
supervisors, technicians analyses
Company’s
Maintenance
Management
Knowledge
records
models
Leadership:
1. Recognition,
2. Empowerment,
3. Interest
OMDEC team participants

Murray Wiseman – LRCM, CBM
specialist

Dr. Daming Lin – Maintenance data
statistician and reliability expert, signal
processing, reliability software, database +
ETL specialist.