How to Achieve

Autonomous Operations
Simon Rogers

Proprietary Information 1
Proprietary Information 2
 Housekeeping

Send all questions during the webinar via the Q&A

Q&A
box.

Polls Take part in our on screen polls.

Download A copy of this presentation can be downloaded

presentation using the ‘green’ Resource List button.

Proprietary Information 3
 Your presenter

Simon Rogers
Vice President
Advanced Solutions and
Business Division

Proprietary Information 4
 Agenda

1 Introduction

2 Value Chain Optimization

3 Dynamic Real-Time Optimization

4 Cognitive Operation

5 Conclusion

Proprietary Information 5
 No safety incidents

Net zero emissions

AUTONOMOUS No unplanned outages

OPERATIONS
VISION Nimble response to market
changes and plan disturbances

A motivated and informed workforce

A culture of profitability

Proprietary Information 6
 Aligning business & financial goals with Digitalization actions

OPERATIONAL EXCELLENCE
BUSINESS GOALS KBC OPX MANAGEMENT SYSTEM FINANCIAL GOALS
MANAGEMENT SYSTEM
C-Suite Key OpX Functional Corporate
Agenda Themes Goals Goals
Strategic Direction
& Leadership Organizational
Operational Market Effectiveness
Effectiveness Effectiveness Performance Management
& Continuous Improvement Maintain
HSSE
License
Excellence

Maximize ROCE at Acceptable Risk

Capital Asset to Operate
Management Optimization
Regulatory
Value Chain & Production Compliance
(Revenue Maximization)
Financial
Risk & Policy Supply/ Demand
Stewardship
Balance
Increase
Revenue
Disruptive Risk Management
Safety & Risk & Governance Availability
Technology
Increase
Margins
Variable Costs
Governance Digitalization Asset Design & Utilities &
Improvement Reduce OPEX
Asset Integrity
(Capital
(Cost Containment) Fixed Costs
Management)

Leadership & Operational

Succession Control Min. Working
Finance… Procurement… Sales… IT… HR…
Audit… Legal… Stakeholder Management Capital Optimize
Individual Capital
Capability Ownership & Max. Investment Employed
Capability Returns

Proprietary Information 7
 Empowering the plant to run, learn, adapt and thrive in
tomorrow’s environment
Closed Loop
Optimization
Procedure
Automation
Business
Closed-Loop
Control
Open-loop Autonomous
Advice-based Operations
Action
Level of Automation

Best
Autonomous
practices Orchestration Dynamic Real-
Semi- Time Optimization
Autonomous Modular
Procedure
Automated Automation
Multivariable
Predictive Control
Manual /
Semi-Automated Integrated
Control and Safety
System

Lifecycle Simulation / Artificial

Plant
Intelligence
Time
Proprietary Information 8
 Actions happen; data is produced and distributed.
What makes the uncertainty?

Uncertainty
Cost of
$$$$
Investment
Planning

Decision Value
Supply
Planning

Production
Planning

Scheduling

Automation

NOW

YEARS MONTHS DAYS HOURS MINUTES SECONDS MINUTES HOURS DAYS MONTHS
Ago Ago Ago Ago Ago Ahead Ahead Ahead Ahead Ahead

Decision-Making Decision Impact

Time Horizon Time Horizon
Proprietary Information 9
 Digital value chain and asset optimization
1 month 1 day Current Time 1 to 15 days 1 to 3 months 1 year 2 to 10 years

Supply Chain Optimizaton Strategy Corporate Business Plans

Fiscal Strategic
Supply Chain Supply Chain Supply Chain Targets Plan
Supply Chain Annual Investment
Status Management Scheduling Planning Planning Planning
Feed and Inventory
Plan
Variance Analysis Production
Planning
Operating
Actual Plan
Production
Production Production
Actual Accounting Scheduling
Operation Model Update
Schedule

Asset Optimization Process Simulation

Profit Improvement Process & Utility Process Feasibility
Debottlenecking
Process Unit Management Optimization Design Study

Operating Targets
Process and Off-sites Automation
Procedural Automation APC

Integration Model Update Enterprise Data IIoT Data

Proprietary Information 10
 Today’s challenges
• Limited range of linear models
Production
• Updating models is time- Accounting
consuming and SME
Corrected
dependent model Reconciliation &
Simulation
Variance Analysis
• Largely manual reconciliation
Corrected
work flow model
Operational Corrected
• Data siloed, difficult to manage Planning data

• Slow recognition of Raw

data
opportunities to open Supply Chain
Crude selection Scheduling
constraints Production
targets
• Control strategies not updated Process Automation &
with operating targets Operating targets Optimization
Movements plan
• Limited integration between Plant
tools

Proprietary Information - 1 MONTH - 1 WEEK NOW 11

 Digital future
• Data driven, automated
identification of optimization
using AI
• Intelligent, automated work
processes
• Automated data management
and integration
• Cloud enabled to facilitate;
• Scalability
• Collaboration
• Support
• Rapid enhancements
• Knowledge management
• Integration
• Visualization
Proprietary Information 12
 Automated
process

Near real-time planning

and scheduling

Proprietary Information 13
 Data driven operation C.1

C.4
Knowledge graph for managing C.3
1 and visualizing supply chain
data
Demand, price and performance
2 forecasting
Reconciliation, variance analysis
3 and opportunity identification
C.2
4 Automation of model updates C.5
Automation of production
5 scheduling
C.6
C,7
6 Asset-wide optimization
7 Cognitive operations

Proprietary Information 14
 C.1 Value chain knowledge graph
Global optimization Material
1
Place 1
Visualize and manage entire Plant 1

supply chain
Connect data silos – Linked Material
Quality
Cost Place 2
Data 3
Plant 2

Add structure and context

Partner

ML, text mining & NLP Material

2
3
Distance Place 3
Cost
Semantic search and AI Partner Time
3 Mode

Automation, classification
Reporting, personalization Business Taxonomy and Ontology – Meta Data

Unstructured ERP Planning Scheduling Production

Data Database Database Database Database

Simulation Database
Proprietary Information 15
 C.3
Reconciliation, variance analysis and opportunity
identification

Planning/Scheduling Optimal Solution Actual Realization

Model

Plan Constraint Set Expected

Planning
Benefit
Optimal Targets
Historical
Information
Schedule Constraint Set
Scheduling
Desired Actions Actual
Benefit
Actual Actions

Finding improvement opportunities automatically

Programs on > 150 sites
Operation
Rigorous Site Model
Plant Profit Planning to Validate
KBC PIP
Base Case Opportunities Proposed
Forecasts Improvements
- Market Conditions
- Site Operating Conditions Opportunity
Engine
Proprietary Information 16
 C.3 Early detection of opportunities
Abnorm
Normal Caution
al

Actual, Simulation and Plan Data

Improvement opportunity index

Time

Identify ‘potential’ opportunities by monitoring indices in real time

Proprietary Information 17
 C.5 Automation of production scheduling

Baseline Crude Refinery Blending Shipping

Update Scheduling Scheduling Scheduling Scheduling

Opening Inventory and Violated constraints Task updates

inventory past task Deviations from plan Nomination updates
discrepancies updates Economics

Explained
AI Classifier
MCTS, Supervised
and Reinforcement
Learning

Proprietary Information 18
 C.6
C,7 Dynamic real-time optimization

• Steady State Model • Continuous profit improvement

• Single Unit Co-Pilot • Model maintenance
CPM
RTO • Hourly

• Asset wide optimization and

scheduling
Petro-SIM
• Multivariable Control VM -SCS • Daily targets
• Every Minute
APC
• Economics, Gains & inferential calcs on
change of feed or operation
Petro-SIM
• Multi-unit dynamic optimization / minute
PACE

• PID Control
• Every Second • PID Control
DCS
DCS • Every second

Proprietary Information 19
 C.6 Simulation synchronised with real-time production
C,7
system

Proprietary Information 20
 C.6
C,7 Plant-wide dynamic optimization

Economics On-Line Petro-SIM and Machine

1/0.5~1hr
Schedule Learning Models
Integration with Petro-SIM
SS Gains MV / CV Limit
Assay / Inferred CVs ON / OFF Status provides inferential
Feed Properties Objective calculations and gain
function
updates
Plant –wide APC and Optimization
Historian Sub Controller 1 1/5 mins Tight integration with APC
Sub Controller 2 ensures better disturbance
3
OPC
rejection and easier
OPC maintenance

DCS 1/sec

Plant

Proprietary Information 21
 C.6
C,7 Non-linear multi-unit optimization
Gas Plant

Unit Light Ends

Light Naphtha H2
ISOM
REF
Heavy Naphtha
HTR
C Kero
D HTR Blending
U
Diesel
HTR

ALK
Atmospheric VDU Gasoil HCR
Residue VGO
FCC
HTR
COKER
Vacuum Residue
Coke

RHDS

Optimization distributed across plant areas defined by intermediate tankage

Proprietary Information 22
 C.6
C,7 Olefins plant-wide dynamic optimization

Petro-SIM Plant-Wide Model

Targets
Model Gains

Petro-SIM Furnace
Model
Inferred Values
Controller Gains Plant-Wide Dynamic Optimizer
Run length

Furnace Furnace Hot Section C1 C2 C3 C4

Multivariable Multivariable Multivariable Multivariable Multivariable Multivariable Multivariable
Control Control Control Control Control Control Control

CVs MVs CVs MVs CVs MVs CVs MVs CVs MVs CVs MVs CVs MVs

Furnace 1 ……. Furnace X

Quench /
CT/DEC1
C2H/DEC2/ C3H/DEC3/
DEC4
Fractionator EE Splitter PP Splitter

Proprietary Information 23
 C.6 Optimal trajectory reduces time to optimum
C,7
operation
Global Optimization by Petro-SIM
Optimum
Profit

Profit

Expected Profit
from
Traditional RTO
Time
(Hours)
Traditional RTO Waits for Steady State

Proprietary Information 24
 Mixed integer non linear utilities optimization
C.6
C,7

Emissions
CLOSED-LOOP OPEN-LOOP
Utilities Optimization signals to Process Simulation advisory reports to
Regulations
control systems operators

EnPIs, Streams
and Equipment Energy
Measurements Calcs. Measurements
Optimum Related
Energy Reports and
System Dashboard
Control System / Historian Operatio
External Utilities Operators
n Report
Contracts
Energy and
Energy and Utilities System Emissions
Hydrogen Fuel Steam Water Electricity Process KPIs
(Utilities Monitoring
Consumers) Energy Use
and Account
(Degrees of freedom in Generation and Distribution)

Proprietary Information 25
 C.6
C,7 Cognitive operations
Steady State Simulator Outputs
Inputs Yields
Feed rate & quality Compositions
Operating Energy
parameters Catalyst activity
Parameters HX coefficients
- Tray Efficiencies
- Calibration Factors
PACE
Feed rate & quality APC/RTO
Yields On-line ML models
Compositions
Fuel flow Process and
model KPIs Knowledge Graph
+ Symbolic AI
Business & process
knowledge Continuous
Best practices learning
Operations Advisor
Yield
Energy

Increase CDU COT

Increase FCC ROT
Increase Kero stripping steam
Proprietary Information 26
 C.6 Cognitive operations require Cloud, IIoT, Big Data and
C,7
AI

Cloud Portal and Digital Twin Cloud and IIoT based real
time collaboration across
supply chain
Digital Twin to validate
and optimize

Connected Experts
Customer Location

Process Simulation performance using

mathematical
programming, first
LP Sub Model principles models and
machine learning
Insight,
recommendations,
Data Centre

reporting and
stewardship by experts
Proprietary Information 27
 To achieve autonomous operations

◼ Integrated value chain optimization

◆ Planning, scheduling & accounting
◆ Simulation, machine learning &
knowledge graphs

◼ Dynamic real-time optimization

◆ Multivariable predictive control
◆ Simulation & machine learning
◼ Cognitive operations
◆ Cloud computing & big data
◆ IIoT & AI

Proprietary Information 28
 Now is the Time
Digitalization
Simon Rogers the scalable application of the digital
Vice President, technologies, and alignment of the
Yokogawa Advanced Solutions organizational capabilities that we believe an
+81 70 4226 0673
energy or chemical process operation should
Simon.Rogers@jp.yokogawa.com
www.kbc.global have and master, with digital information at
the core, in order to achieve excellence

Proprietary Information 29
 Q&A

Q&A Send your questions via the Q&A box.

Download A copy of this presentation can be downloaded

presentation using the ‘green’ Resource List button.

Take our short survey to give us your feedback on

Survey
today’s webinar.

Proprietary Information 30