4. WE ARE IN AN UNPRECEDENTED
PERIOD OF TECHNOLOGY INNOVATION
1 MAINFRAME
2 CLIENT-SERVER
AND PCS
3 WEB 1.0
ECOMMERCE
4 WEB 2.0,
CLOUD, MOBILE
BIG DATA, ANALYTICS,
VISUALIZATION5
IOT AND
SMART MACHINE6
ARTIFICIAL
INTELLIGENCE7
QUANTUM
COMPUTING8
TODAY
1950 1960 1980 1990 20201970 2000 2010 2030
1950 Turing Test
2005: Web 2.0 Quantum
1964: System/360
Server/Host
1969:
ARPANET
1990: System/390
1991: Public Internet
1994: Amazon
1977: PC
1999: Salesforce.com
2006: AWS
2008: iPhone
1997: Big Data
Public Cloud Mainstream
2010: Sales of PC Peak
2010:
Self-driving Car
AI
2014, IDC:
4.4 Zettabytes of Data
1972: SAP
1999: IoT, M2M
5. WHAT IS ARTIFICIAL
INTELLIGENCE (AI)?
It is the single biggest technology
revolution the world has ever seen.
SENSOR
PROCESSING
DEEP
LEARNING
ROBOTIC
PROCESS
AUTOMATION
EXPERT
SYSTEMS
INFERENCE
ENGINES
MACHINE
LEARNING
COMPUTER
VISION
KNOWLEDGE
REPRESENTATION
SENSE
COMPREHEND
ACT
LEARN
7. AI powered service that answers
customers’ common lubricant
questions in seconds
AI robots for ocean exploration
to improve natural seep
detection capabilities
AI to accurately and speedily
predict quality of gas products
By 2020, 85% of customer
Interactions will be managed
without a human
8. AI IS DRIVING ECONOMIC VALUE
ACROSS THE VALUE CHAIN
Through three channels of AI-led growth that drive
increased productivity, satisfaction, and value…
R&D /
LABORATORY
PLANTS, ASSETS,
MANUFACTURING & OPERATIONS
SUPPLY CHAIN
& INVENTORY
COMMERCIAL
& SALES
ENTERPRISE
FUNCTIONS
INTELLIGENT AUTOMATION
Creates growth through a set of features enhancing traditional automation solutions.
LABOR & CAPITAL AUGMENTATION
Growth will come from enabling resources to be used much more effectively and valuably
INNOVATION & DIFFUSION
Ability to propel innovations as AI diffuses through the economy.
12. APPLICATIONS
• Automobile parts
• LCD displays
• Extrusion sheets, optical
lenses
• Tableware
• Swimming pools and water
feature projects
COGNITIVE PLANT OPERATIONS
APPLYING AI TO
IMPROVE QUALITY…
...by consistently reducing defects
through optimal plant conditions
...from data capture to actions
taken by operations, maintenance
& reliability team
Automotive Parts Construction: Stadium Roof
PRODUCT
13. CHALLENGES TODAY
DRYER
Spray Water
Overflow Water
CONVEYOR WATER
Lower
Feed Roll
EXTRUDER
PELLETIZER
MMA
Powder
Additives
1
FEED
2
EXTRUDING
3
STRAND
COOLING
4
PELLETIZING
5
PELLET
COOLING
6
DRYING
Hopper
Thermocouple
NozzleBarrelScrew Heaters
Cutting
Rotor
Cutting
Blade
Upper
Feed Roll
CONVEYOR
Operators follow a trial and
error approach towards
reducing defects
Cutting rotor changed without
clear reason or justification
Golden recipes looking at 70
parameters out of which operators
are currently adjusting 7 parameters
14. APPROACH
Engineers /
Supervisors NLP,
Intelligent Search
In History
Data Science To
Model The Plant
Form-less Interaction
With Virtual Assistant
2016
2015
2014
2013
2012
2011
0 20 40
0
2
4
6
0
2
4
6
0
2
4
6
0
2
4
6
0
2
4
6
0
2
4
6
week
wday
crack_category
Average
Bad
Good
Calendar Plot for MH 100 Crack
Categorically Bad > 15 Defects
Categorically Good < 10 Defects
Categorically Average 10 – 15 Defects
TI2
403.P
V_m
ed
75
80
85
90
SC2401.PV_med
138
139
140
141
142
143
fittedvalue
8
10
12
14
16
18
Good
Bad
No. of defects
Good quality operating zone
15. HELPING OPERATORS LEVERAGE DATA
SCIENCE & AI TO FIND BEST RECIPE
SMART
Predict current
cutter condition &
degradation levels
Predict optimal setting for
adjustable parameters
in near real time
12 variables that
impact quality
and defects
Automatically identify
past load changes,
quality measurements
Apply best in class
settings achieved in the
past to current situationsAFTER
Sub-optimal change
of cutter
Currently adjusting
only 7 variables
Trial and Error
Approach taking days
Memory / experience
driven actions
BEFORE Hand written
operator logs
WISE
17. OBJECTIVE: Use Artificial Intelligence to optimize the operating conditions of a cracked gas compressor for maximizing production benefits.
APPROACH
CHALLENGES
FEEDSTOCK
STORAGE
PROCESS OPTIMIZATION
1
SCOPE
DEFINITION
2
DATA
GATHERING
& ANALYSIS
3
MODEL
DEVELOPMENT
& EVALUATION
4
DESIGN
THINKING
& CHATBOT
CREATION
5
VALUE
EXPANSION
FURNANCE
SET UP
QUENCHING 5 STAGE COMPRESSOR DISTILLATION
INPUT SCOPE INPUT
19. DARK DATADARK DATA: LOOKING AT THINGS DIFFERENTLY
TECHNOLOGY CAN HELP HUMANS SEE MORE IN WHAT THEY HAVE.
USING TECHNOLOGY TO BYPASS HUMAN PROCESSING CONSTRAINTS, E.G.
NLP, COMPUTER VISION, VIDEO ANALYTICS, DEEP LEARNING…
INSIGHTS
FROM
UNTAPPED
DATA
Email
Correspondences
Notes/
Presentations
Survey Data
Document
Versions
REPOSITORIES
Video
Surveillance
Access Logs
Audit Files
Internet Logs
MONITORING
Employee Data
Purchasing
History
Usage LogsCustomer/Account
Information
Financial
Statements
OPERATIONS/ BI
STRUCTURED +
UNSTRUCTURED
LARGE VOLUMES
REAL-TIME +
HISTORICAL Maintenance
Records
20. INSERT VIDEO SCREENSHOT HERE
Natural Language
Processing (NLP)
capabilities enabling smarter
business decisions
SMART DATA
EXTRACTOR
UNSTRUCTURED
TO STRUCTURED
DATA
21. IN SUMMARY…
AI IS ALREADY QUITE
PERVASIVE - THE TIME
TO ACT IS NOW!
AI DOESN’T REPLACE
BUT RATHER AUGMENTS
THE WORKFORCE
HARNESSING AI WILL
ONLY HELP BOOST
YOUR BUSINESS
CURRENT TRENDS ARE
MAKING AI ACCESSIBLE AND
EASY TO IMPLEMENT TODAY
23. WHAT VALUE DOES AI POTENTIALLY DRIVE?
In 2035, AI will …..
DOUBLE ECONOMIC GROWTH
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
US Finland UK Sweden Netherlands Austria Germany France Japan Belgium Spain Italy
Baseline AI Steady State
Real GVA* Growth in 2035
24. WHAT VALUE DOES AI POTENTIALLY DRIVE?
In 2035, AI will …..
Source: Accenture and Frontier Economics
29%
30%
34%
35%
36%
37%
11%
12%
17%
20%
25%
27%
LABOUR PRODUCTIVITY BY 40%
Help People be More Productive and Increase
Sweden
Finland
United States
Japan
Austria
Germany
Netherlands
United Kingdom
France
Belgium
Italy
‘Spain
Editor's Notes
The backdrop to all of this is that we are in the midst of a massive amount of technological change.
We are moving from scarcity to abundance.
Cost per million transactions was $222 in 1992 to 6 cents in 2012.
Cost for sequencing a genome in 2001 was $100 million … in 2015, it was $1,000 (https://www.genome.gov/images/content/costpergenome2015_4.jpg)
Cost for hard drive storage in 1980 $437,000 per gig … now it is .019 cents per gig 2016 (http://www.statisticbrain.com/average-cost-of-hard-drive-storage/)
An industrial 3D printer in 1987 was $800,000 … in 2012, it was $15,000 … and in 2013
In 2007 a personal 3D printer was $4,000 and in 2013, it was about $1,000 (and dropping) (Accenture 3DP Article)
Intelligent Automation: Creates growth through a set of features enhancing traditional automation solutions.
Ability to automate complex physical world tasks that require adaptability and agility
Ability to learn by experience and improve, enabled by repeatability at scale
Based on software robotics + machine learning/deep learning to respond autonomously for it’s defined task. Examples: Customer Service (Answering questions, providing recommendations), Insurance Claims (how to process, adapts the activity based on performance, can be proactive to handle to minimize exceptions/mistakes). It’s base on RPA (robotic process automation) capabilities and then layered with machine vision, self-learning, and adaptive capabilities to enhance the process accuracy, flexibility, and autonomy.
Predictive algorithms for supply-chain
Automate planning and scheduling of raw material purchase
Demand forecasting for more effective pricing strategies
Labor and Capital Augmentation: Growth will come from enabling resources to be used much more effectively and valuably
Enable humans to focus on parts of their role that add the most value
Improve capital efficiency—a crucial factor in Industries where it represents a large sunk cost
Using Machine Learning capabilities to provide proactive recommendations, advice, and personalization that enhances performance/productivity. This includes inference engines, predictions, or expert systems. Also AI can be used to optimize the use of raw materials, assets, and capital for an organization – either through analysis and actions or to recommend optimal solutions to buyers, procurement, and operations leads.
Innovation Diffusion: New products that are based/rely on AI. This includes autonomous cars, Alexa, and Loon (Alphabet’s system to provide global internet).
Optimize operational efficiency
Intelligent robots – Reduced worker risks, improve productivity and cost-effectiveness
Virtual assistance – Navigates database, and processes general inquiries using natural language
InNOVATION & DiFFUSION Ability to propel innovations as AI diffuses through the economy.
Innovation begets innovation, the potential impact of an AI solution expands to new products/industries
Opens new business models and opportunities
Enhancing Product innovation and integrity
Modeling oil-refining strategies
Compliant Product Lifecycle Management
CHALLENGES
Bottleneck on turbine torque is limiting the compressor capacity to produce more
Simulations based first principle is tedious and does not holistically generate insights that are visible using big data
Reliability concerns/fear of increasing turbine speed
Data Science Led Approach
Deploy Data Science coupled with process expertise to derive valuable insights from a tight operating window
Visualize analytical outputs, delivering insights to operators upfront, ready for consumption
Insights on different operating states / trends and key drivers of Ethylene yield
Insights on optimization / improved efficiency of the compressor and production management at given energy levels (set points that need to be adjusted towards improved operations)
CHALLENGES
Bottleneck on turbine torque is limiting the compressor capacity to produce more
Simulations based first principle is tedious and does not holistically generate insights that are visible using big data
Reliability concerns/fear of increasing turbine speed
Data Science Led Approach
Deploy Data Science coupled with process expertise to derive valuable insights from a tight operating window
Visualize analytical outputs, delivering insights to operators upfront, ready for consumption
Insights on different operating states / trends and key drivers of Ethylene yield
Insights on optimization / improved efficiency of the compressor and production management at given energy levels (set points that need to be adjusted towards improved operations)
CHALLENGES
Bottleneck on turbine torque is limiting the compressor capacity to produce more
Simulations based first principle is tedious and does not holistically generate insights that are visible using big data
Reliability concerns/fear of increasing turbine speed
Data Science Led Approach
Deploy Data Science coupled with process expertise to derive valuable insights from a tight operating window
Visualize analytical outputs, delivering insights to operators upfront, ready for consumption
Insights on different operating states / trends and key drivers of Ethylene yield
Insights on optimization / improved efficiency of the compressor and production management at given energy levels (set points that need to be adjusted towards improved operations)
Intelligent Automation: Based on software robotics + machine learning/deep learning to respond autonomously for it’s defined task. Examples: Customer Service (Answering questions, providing recommendations), Insurance Claims (how to process, adapts the activity based on performance, can be proactive to handle to minimize exceptions/mistakes). It’s base on RPA (robotic process automation) capabilities and then layered with machine vision, self-learning, and adaptive capabilities to enhance the process accuracy, flexibility, and autonomy.
Labor and Capital Augmentation: Using Machine Learning capabilities to provide proactive recommendations, advice, and personalization that enhances performance/productivity. This includes inference engines, predictions, or expert systems. Also AI can be used to optimize the use of raw materials, assets, and capital for an organization – either through analysis and actions or to recommend optimal solutions to buyers, procurement, and operations leads.
Innovation Diffusion: New products that are based/rely on AI. This includes autonomous cars, Alexa, and Loon (Alphabet’s system to provide global internet).
Intelligent Automation: Based on software robotics + machine learning/deep learning to respond autonomously for it’s defined task. Examples: Customer Service (Answering questions, providing recommendations), Insurance Claims (how to process, adapts the activity based on performance, can be proactive to handle to minimize exceptions/mistakes). It’s base on RPA (robotic process automation) capabilities and then layered with machine vision, self-learning, and adaptive capabilities to enhance the process accuracy, flexibility, and autonomy.
Labor and Capital Augmentation: Using Machine Learning capabilities to provide proactive recommendations, advice, and personalization that enhances performance/productivity. This includes inference engines, predictions, or expert systems. Also AI can be used to optimize the use of raw materials, assets, and capital for an organization – either through analysis and actions or to recommend optimal solutions to buyers, procurement, and operations leads.
Innovation Diffusion: New products that are based/rely on AI. This includes autonomous cars, Alexa, and Loon (Alphabet’s system to provide global internet).