PROGNOSTICS

THE KEY TO PREDICTIVE MAINTENANCE

@senseyeIO
Me
• BEng Digital Systems Engineer

• Background in aerospace & defence and large

scale wireless sensing
• Software Verification & Validation
• Wireless sensor networks (EMMON)
• Health & Usage Monitoring Systems / Condition
Monitoring Systems
problem: downtime

300%
The amount by which Total
$2.5m
The cost per hour of
Downtime Cost is usually downtime in automotive
underestimated manufacturing

24%
Percent of total
90%
manufacturing cost Amount of maintenance work
attributed to categorised as ‘crisis work’
downtime to fix breakdowns

Downtime is expensive
I D E A L P R E V E N TAT I V E M A I N T E N A N C E :
OPERATING OPERATING OPERATING OPERATING

MTBF MTBF MTBF MTBF

PLANNED PLANNED PLANNED

MAINTENANCE MAINTENANCE MAINTENANCE

R E A L - W O R L D P R E V E N TAT I V E M A I N T E N A N C E :
OPERATING OPERATING OPERATING OPERATING

MTBF MTBF MTBF MTBF

PLANNED PLANNED PLANNED

MAINTENANCE MAINTENANCE MAINTENANCE

UNPLANNED UNPLANNED UNPLANNED

BREAKDOWN BREAKDOWNS BREAKDOWN
(panic) (double panic) (sigh)
old: new

PREVENTATIVE PREDICTIVE
Known costs + surprise items Exact knowledge of machine condition
Replacement of good parts Parts only replaced when deeded
Potential for unintended damage Fewer surprises
Inefficient Reduced operational costs
No idea of actual machine condition Reduced downtime

REACTIVE SCALABLE
IDEAL PREDICTIVE MAINTENANCE:
OPERATING

! ! !
OPPORTUNISTIC
MAINTENANCE

CONDITION
WARNING

REAL-WORLD PREDICTIVE MAINTENANCE:

OPERATING OPERATING

! !
OPPORTUNISTIC PLANNED
MAINTENANCE MAINTENANCE
HOW LONG?
CONDITION
WARNING
 Is not a
dirty word

Understanding when a
PROGNOSTICS machine will stop being
able to work
 prognostics: evolution
Health and Usage Prognostics research
Monitoring for industrial
Systems (HUMS) applications grows
introduced

NOW
1986 1999 2009

1990 2006

British Senseye releases

international The birth of Prognostic Health automated
helicopters crash modern industrial Management condition
automation Society is formed monitoring &
systems prognostics
the birth: condition monitoring

Does it sound funny?

CONDITION
MONITORING The good:
It can be very accurate
Improves safety
Identifies failures
Shows current health

The bad:
It’s expensive
It’s slow
Tr a i n i n g r e q u i r e d
Needs specialist interpretation
Can be slow to show benefits
Cannot scale
T h e U G LY
Diagnostics provides Current State
what is: prognostics

Forecast failure Models of failures Beyond Diagnostics

• Forecast of when • Predict complex failure • Condition Monitoring is only
functional failure will conditions part of the task
occur
• Use of machine context • For action, foresight is required
• Predict future state
• Based on data characteristics of
failures

Prognostics provides Remaining Useful Life

prognostics: remaining useful life
PERFORMANCE
/CONDITION
Normal operation
band

Anomaly detection threshold

Diagnosis threshold

Variation in RUL

Functional failure threshold

OPERATING TIME
Predictive Prognosis
Diagnostics

Senseye covers the full timeline

Prognostics: Trajectory Modelling
(instance based)
Example RUL trajectories for many
failures over time of similar machine
type
Note: not all failures are equal!
Scaling the Diagnostics Engineering Role

Human resource vs Connected devices

Human resources don’t scale

Automate the Diagnostics Engineering Role

it’s: automated

&
C O N D I T I O N M O N I TO R I N G PROGNOSTICS
PROBABILITY OF RISK DEGRADATION
ANOMALY DETECTION AUTO OPTIMISATION
FORECASING FORECASTING

AUTO CONDITION REMAINING USEFUL

AUTO TRAINING
EXTRACTION LIFE

STATE AND USAGE AUTOMATED

DETECTION THRESHOLDING

“How is my machinery behaving right “For how long will my machinery continue
now?” to work?”
key: benefits

10 – 40% 45 – 55%
Lower Increased
maintenance maintainer
costs productivity

30 - 50% 85%
Increase in
Reduction in maintenance
downtime accuracy

Avoid downtime and save money

 Predict the
Unpredictable

Reduce unplanned downtime and maximize

OEE with Senseye, the ultimate cloud-based
predictive maintenance application
technology: convergence

Condition monitoring

Industry
4.0 / Prognostic
Industrial s
IoT

Machine learning / AI
 Remaining Useful Life
Understand for how long your machinery is likely
Any & many to perform

Any connected machine is Improved predictive maintenance planning

supported, big or small, from 10
to 10,000
Scale effortlessly “Motor has a
70% risk of failing
within two weeks”

Web based
Condition Monitoring Nothing to install or maintain,
Automated condition monitoring insights, only powered by the cloud and
relevant information is highlighted. accessible on all devices.

Takes the pain out of asset management No infrastructure to support

removing: downtime
Amount of uptime
Panicked repairs

On Time and In Full

(OTIF) deliveries

Overall Equipment Costs

Effectiveness (OEE)
“
The factory of the future will have only two employees, a
man and a dog.

The man will be there to feed the dog.

The dog will be there to keep the man from touching the
equipment.
- Warren Bennis
 @senseyeIO

[email protected]

THANK YOU +44 (0) 845 838 8615 (UK)

+44 (0) 7818 068 861 (Alex)