The industry is reeling under the explosion of data generated by smart sensors, motors, actuators, machines, and other “things”. With the pace at which production is happening currently, the last straw would be an asset breakdown. Statistics show that the automotive industry deals with an alarming 800 hours of downtime every month. The cost of such downtime is a staggering US$22,000 per minute, or US$12.6 million a month.
Additionally, data shows that 20% of these breakdowns are common or predictable and that a majority – a shocking 80% – of them are seemingly random instances and cannot be predicted.
According to McKinsey, the Industrial IoT (IIoT) market is worth $11 trillion, and predictive maintenance solutions can help companies save $630 billion over the next 15 years. So, how can manufacturers tap these savings and benefits?
Learn how manufacturer and suppliers can experience the power of Cognitive Predictive Maintenance (CPdM) to avoid unplanned downtimes and drive greater efficiencies.
4. DataRPM Overview: Silicon Valley VCs to NASDAQ Acquisition
A Cognitive
Platform for
Predictive
Maintenance
Prescriptive
Analytics for the
Industrial IOT
using
50+Employees
[42+ Engineers]
2012
Focus Established Our Team Team Composition VC / M&A
HQ
Silicon
Valley,
CA
R&D
Bangalore
India
SATELLITE
Fairfax,
VA
SATELLITE
Belfast
UK
Acquired March 2017 by
(now an independent subsidiary of
NASDAQ-listed Software Leader: PRGS)
Now fueled by Progress’ Global Reach:
1,700+ Independent Software Vendors
80,000+ Enterprise Customers in 175 Countries
2,000,000+ Developers
6,000,000 Users of Progress-enabled Applications
9. 9
The Paradigm Shift Towards Being Predictive
Owing to a dynamic market scenario, companies are increasingly applying analytics for equipment maintenance and asset management
purposes, as they offer a quick turnover
Traditional Approach Diagnostic or Reactive Approach Predictive or Proactive Approach
Drawbacks of traditional and reactive approaches: High costs associated with equipment downtime and
decreasing efficiency of equipment to impact utilisation and production rates
UnplannedMaintenacne
• Typically, manufacturing
facilities would receive a
weekly or monthly report on
the production line and
individual equipment
performance
• If a problem is identified, a
field service technician would
be deployed and would be
able to use the logbook
maintained by previous
technician in conjunction with
maintenance orders in
enterprise system to identify
the last maintenance related
activity performed
PlannedMaintenance
• Data is collected from assets
and benchmarked with
historical data using simple
statistical tools to evaluate
the root cause of failure
• The company will come up
with a strategy to identify
similar instances with similar
asset classes and set
accurate tolerance limits to
minimise future failure
• System efficiency is usually
improved post breakdown of
specific equipment on the
production line
ProactiveMaintenance
• Complex statistical
algorithms and machine
learning techniques are used
on real-time data to predict
failures well in advance
• Advent of prescriptive
analytics identifies the impact
of equipment failure on the
surrounding environment
(e.g., process flow)
• By identifying anomalies
early, companies can
efficiently deploy field service
personnel to streamline
maintenance activities
10. 10
Using state-of-the-art open source platforms that
facilitate the development of custom apps across
multiple industries
Push towards new supplier business models, where
pricing is governed by outcome-based and
consumption-driven methodologies
Leveraging asset connectivity and Big Data to derive
insights from machines/assets/equipment to improve
overall process efficiency
Supplier focus shifting from vertical applications to
more horizontal applications across manufacturing
segments
Transition of manufacturing business models from
traditional asset supply to long-term service driven
engagement
Five Strategic Shifts of Industry 4.0
The technological advancements of Industry 4.0 aimed towards real-time performance monitoring and mass customisation is
shifting the strategic focus of global businesses
Industry 4.0 (Smart Factory): The term Industry 4.0 that originated in Germany is a new manufacturing framework built
around the concepts of cyber-physical production, Internet of Things, enterprise mobility, new service models, and cloud
computing technologies
Strategic Shift 1: Asset à Services Strategic Shift 2: Vertical à Horizontal
Strategic Shift 4: Subscription à Consumption Strategic Shift 5: Closed Loop à Open Source
Strategic Shift 3: Machine à Data Driven
Predictive
Analytics
Data
Management
11. 11
Workflow of Cognitive Predictive Maintenance With Meta Learning
Cognitive Predictive Maintenance access data from connected equipment to integrate with existing service knowledge to
understand and address the issue with an intent to improve quality and service time
Sensor
(Batch Time Series Data)
Feature Engineering Anomaly Detection
Labeled Training Data Prediction Modeling Production
Connectors Feature Engineering Process Segmentaion
Process
Influencing Factors Identification Prediction Process API Framework + Scoring Process +
Recommendations + Dashboard
Meta Learning Application
Prediction
accuracy
increase at an
average of 300%
Results
delivered
almost 30 X
faster
Average of 75%
reduction in
breakdown
• Specifically designed to handle the challenges of predictive maintenance for IIoT
• Cognitively automate the data science process at mass scale
• Utilize Meta-Machine-Learning
• Operationalize the best ensembles and continually modify in-line & real-time
Key Characteristic
12. 12
Advanced Machine Learning – A Proactive Approach
Predictive and prescriptive analytics to expand at a CAGR of 56.9%
Market Size
$1.2 billion
2016
Metrics
Descriptive &
Diagnostic
Predictive &
Prescriptive
Revenue
(2016)
$0.83 billion $0.37 billion
Revenue
(2021)
$1.89 billion $3.51 billion
CAGR 17.9% 56.9%
Significant change in the
percentage revenue split by product
segment over the forecast period
Market Size
$5.40 billion
2021
Predictive & Prescriptive
Descriptive & Diagnostic
20%
80%
65%
35%
Descriptive & Diagnostic Analytics Predictive & Prescriptive Analytics
• Uses simple statistical tools to pin point the reason for
the failure
• While the system efficiency may be improved post
failure, there are significant costs incurred due to
equipment downtime.
• Uses complex statistical algorithms and machine learning
techniques to benchmark historical data with real-time
sensor data
• Numerous benefits with regard to cost, process efficiency,
and even equipment self-learning from surrounding
environments (prescriptive)
13. 13
Key Workforce Requirement – Predictive Maintenance
Apart from potential opportunities in data management, there is a growing requirement for engineering roles in device
communication and quantum data storage technology
Build Validate &
Deploy Model
Optimize &
Dispatch Service
Technicitions
Evaluate & Monitor
Results
Data Preparation &
Exploration
Sensor Controller
Data
Historian
Life Cycle of Predictive
AnalyticsBusiness Analyst
Responsible for data exploration, reporting, and
visualisation
Data Statistician
Responsible for descriptive data segmentation and
predictive modelling
IT Systems/Management
Responsible for data preparation, model building,
validation, and deployment
Manager
Responsible for evaluating resulting and making
informed decision
Workforce Requirements:
As predictive analytics gains precedence across
industrial environments, there is a burgeoning need for
personnel with hybrid skills
14. 14
Key Industry Challenges
Plant maintenance decisions incur high cost and mean-time for equipment repair which ultimately affects the profitability
Key Industry Challenges
Interpretation of
collected data
Data that is collected, stored, and transmitted from
machines through sensors require a proper means to
decipher and understand the factors that cause
equipment to work a certain way.
Inaccuracy in
calculating
downtime cost –
Tangible &
Intangible
Average plant downtime costs the automotive industry over $1
million a month. Tangible costs are easy to determine by simply
considering the difference between planned and actual
operating time while, mostly intangible costs like stress on
equipment and workers as a department attempts to catch up
are not taken into account
Lacking focus on
key elements of
competitiveness
Plant downtime, not just affect the cost but also will have an
serious impact on quality and lead time. While on the other
hand, manufacturers look at maintenance as a strategic function
and not as an insignificant deviation from mainstream
manufacturing.
Machine First Approach
Meta Learning Capability
Accuracy & Speed in
Predicting Failures
Cost Saving Potential
Industry Best
Practices
15. 15
DataRPM Receives Frost & Sullivan Technology Leadership Award
(North American Predictive Maintenance in Automotive Manufacturing)
DataRPM’s predictive
maintenance tool aids
asset-intensive
industries gain a
competitive advantage
by transitioning from
preventive to predictive
maintenance
16. 16
Success Stories – Case Examples
DataRPM’s platform increases prediction accuracy by 300% which ultimately results in faster delivery and cost saving
Business Challenge Solution Impact
Predicted factors for increasing manufacturing efficiency for a prominent car manufacturer in UK
To identify and predict which external
factors affect machine efficiency as a key
performance indicator and how
DataRPM’s CPdM Platform identified
segments with High OEE and low power
consumption to recommend prescriptions
for achieving higher machine efficiency
• Delivered hourly, 3 hourly and daily roll-
ups of production log with weather, traffic,
electrical, temperature
• Machine generated insights based on the
data itself for users who didnt know where
to start their analysis resulted in 3%
improvement in operational performance
Fueled warranty claims and risk transformation for a global automotive
A leading car manufacturer faced the
issue of reduction in customer
satisfaction post car sales due to
frequent warranty claim
Identified 92% Car Equipment failures in
advance and also the reason for each
failure using an automated predictive
model through DataRPM’s CPdM Platform
for ‘Car Part Failure Prediction
• Shoot-up of customer satisfaction rating
as seen the last survey by the
manufacturer
• Improvement in NPS score for the
manufacturer from 28% to 42% in 6
months about the service experienced