Upload
dangnguyet
View
232
Download
1
Embed Size (px)
Citation preview
The Journey to Asset Performance Excellence in Petroleum
Refining & Petrochemicals
Presented by :
Dr. Martin A. Turk
Schneider Electric
Agenda
> Business and Technology Drivers for Asset Performance Optimization
> Digitizing the Value Chain for Asset Performance Optimization
> The Evolution of Maintenance Practices
> The Power of Asset Performance Optimization
> Applications of Predictive Analytics in Petroleum Refining & Petrochemicals
> Value Proposition and Case Studies
Business and Technology Drivers
Market Environment Technology TrendsCustomer Imperatives
Operational
efficiency to drive profitability
Performance visibility and
enhanced decision support
Asset reliability
and availability
Capital expenditure constraints
Workforce evolution
and capability
Commodity Prices
and Oversupply
Compensation
and Consolidation
Environment,
Quality Safety
Regulations
Geopolitical
Uncertainties
Generation
Shift
Cloud
IIoT
Industry Specialization
Mobility
Big Data /
Predictive
Analytics
Empower People by Digitizing the Value ChainClosing the gap between IT & OT to achieve maximum value for your assets
How do I design and commission my assets at the lowest possible cost?
Process Engineering• Process Design & Simulation
• Training Simulators
How can I produce safely and profitably and meet regulatory norms?
Operations Management• Real-time Optimization
• Advanced Process Control
How do I ensure availability and reliability of assets?
Asset Management
• Asset Health & Performance Management
• Mobile Workforce Management
How can I enable better decision making?
Information Management
• Enterprise Historian
• Intelligence & Predictive Analytics
How can I monitor and control operations better?
Operations Control
• HMI
• Supervisory Control
• DCS / PLC
Pla
tform
sA
dva
nce
d A
pp
lica
tions
Steps to Achieve Enterprise Asset Performance Management (EAPM)
> Connect: To Plant-Wide Systems and Sources (IIoT)
> Collect: Capture high-fidelity process, production and equipment information on premise or in the cloud
> Analyze: Apply machine learning, advanced pattern recognition and rules-based logic for asset monitoring
> Act: Enable workforce with full EAPM complement, mobility and enterprise collaboration of people, processes and equipment
CONNECT
EAPM ‒ Typical Systems and Applications Utilized
Workflow
Collaboration
Enterprise Asset
Management
DCS, PLC
SCADA
Sensors and Other
Smart Devices
Online Process Data
Non-Instrumented
Data
Production
Information
Machine Learning
Rules
Advanced Pattern
Recognition
ST
RA
TE
GIZ
E
OPTIMIZE
CONNECT
Organizations are Evolving Beyond Traditional Maintenance Practices to Become Predictive
It’s a Journey
ARC studies show that only 18% of asset failures are age-related. Based on
these data, preventive maintenance provides a benefit for just 18% of assets,
such that monitoring for predictive maintenance is a recommended option for
the rest. www.arcweb.com/Lists/Posts/Post.aspx?ID=260
18%
82%
Failure PatternsAge-related failure Random failure
Reactive and
Preventive Programs
Predictive Technology
for Early Warnings
Transform Data Into Actionable InsightsEAPM bridges the IT/OT gap and contextualizes the increasing amount of industrial Big Data
Mobile Workforce Enablement
> Consolidate disparate data sources
> Bridge the IT/OT gap
> Make data available throughout the enterprise with advanced visualization tools, web solutions and APIs
> Turn data into actionable insights
> Enable cultural change and enforcement of best business practices
> Improve plant reliability, safety and operational profitability
> Ensure that workers execute all field tasks required to achieve reliable operations and cost-effective maintenance
Consolidates high-fidelity operational and asset-health data to
improve decision making
Enterprise Data Historian
Mobile workforce enablement ensures best business practices
Analytics Engines are the Heart of the Predictive Maintenance ProcessWhat Technology is Best for Monitoring Petroleum Refinery and Petrochemical Assets?
Analytic Technologies
> Artificial neural networks
> Clustering
> Decision tree learning
> Deep learning
> Fuzzy logic
> Machine learning
> Advanced pattern recognition
> Others...
Analytics Classifications
> Operational/Process Analytics
> Can be empirical (data driven) and/or theoretical (physics
driven)
> Improve efficiency and output of plant
> Business Analytics
> Typically not real time
> Improve planning associated with financial investments for
the business
> Asset Analytics
> Can be periodic and near real time; mostly empirical in
nature
> Greater Return on Assets due to improvement in equipment
reliability and reduction in unplanned outages
Predictive Analytics for the Monitoring of Asset Health
Software based modeling of equipment using
advanced pattern recognition for conditions which are
not definable by known rules:
> Uses historical data to describe how a piece of
equipment normally operates and builds a model
> Continuously monitors behavior in real-time
> Alerts when the operation differs from the historical
norm
> Provides early warning detection of equipment
problems
> Uses advanced analysis capabilities including
problem identification
Traditional Approach to the Monitoring of Asset Performance
Plot-0
4/6/2008 4:39:21.791 PM 4/9/2008 4:39:21.791 PM3.00 days
60
62
64
66
68
70
72
74
76
58
78
96
108
86
102
107
113
94
102
46
64
0
14
110
170
52
70
25
75> Look at 1,000s of trends and 10,000s of
data points
> Cannot identify every failure pattern
> How do you know if asset data trend is normal
for current conditions?
> Look at trends only when the DCS or Asset
Management System sends an alarm
> Causes include production rate changes, trips,
asset failures because equipment damage has
already occurred it’s too late!!!
Example of Early Warning Notification
Subtle changes are difficult to identify
Actual Value
- Predicted Value
Residual
Ou
tbo
ard
Be
arin
g T
em
p (
°F)
Date and Time
First Pattern Recognition Alarm
Another Example of an Early Warning NotificationLP Rotor ‒ L-0 Blade Issue
> Unit was started after an outage and there was a vibration
step change on one of the LP turbines (but, vibration levels
were well below the alarm level)
> Engineering and the plant were notified
> Vibration data was collected and unit was retired for an
inspection
> Bolts on lower half of flow sleeve had broke and flow sleeve
contacted L-0 blades
> Upper half of flow sleeve was no longer supported by lower
half
> Although there was minor damage to the LP blades, the plant
avoided damaging multiple stages of blades, packing, and
diaphragms if there had been a severe blade liberation
> Estimated avoided cost = $4.1M!!!
Outage
Condition-Based Monitoring
When is it Used
How is it Used
> When the condition is known and definable using rule based logic
> Maintenance rule is fixed and does not significantly change based
on equipment loading, ambient or operational conditions
> Intelligent integration of plant process data with EAM systems
> Industrial asset condition monitoring
> Monitoring of pumps, motors, generators and any other industrial
process equipment (runtime, bearing temperature, volume flow,
vibrations, etc.)
> Equipment health and performance problem discovery and actions
Results Achieved
> Improve labor utilization by 10-20%
> Reduce unplanned downtime by 15-25%
> Increase planned work by 20%
Automated Model-Based Performance Monitoring
Automated Model-Based Performance Monitoring
> Robust and accurate on-line inferred
energy efficiencies and “clean”
equipment performance
> Most accurate corrections of drifting
instruments
> Better inference for missing instruments
Opportunities Abound for Delivering Value from Predictive Analytics in Petroleum Refineries and Petrochemical Plants
Petroleum Refineries
> Rotating Equipment:> Hydrogen Compressors
> Wet Gas Compressors
> Air Blowers
> Steam Turbines
> Gas Turbines
> HP-MP Turboexpanders
> MP-LP Turboexpanders
> Boiler Feedwater Pumps
> Others…
> Fixed Assets:> Furnaces/Fired Heaters
> Heat Exchangers
> Reactors
> Electric Systems (Transformers, Breakers, Switches, etc.)
Ethylene Plants
> Rotating Equipment:> Cracked Gas Compressors
> Refrigeration Compressors
> Boiler Feedwater Pumps
> Others…
> Fixed Assets:> Cracking Furnaces
> Heat Exchangers
> Reactors
> Electric Systems (Transformers, Breakers, Switches, etc.)
US Department of Energy: The Value of Predictive Maintenance
> Reduction in maintenance costs (25% to 30%)
> Elimination of breakdowns (70% to 75%)
> Reduction in downtime (35% to 45%)
> Increase in production (20% to 25%)
On average, predictive maintenance can help
industrial process companies in the following ways:
Value Delivered with Enterprise Asset Performance Management
Maximize return on assets by:
> Improving asset performance
> Increasing reliability and reducing
unscheduled downtime
> Increasing asset utilization and
extending equipment life
> Reducing operations and
maintenance costs
Case Study: International oil & gas company deployed a standardized mobile Decision Support System
Closing the Loop
After a successful pilot, the company
selected a mobile operator rounds and
decision support system to collect non-
instrumented data and standardize work
processes. The mobile workforce
enablement tool allows the company to
achieve significant savings and reduce costs
year on year.
Results:
> Standardized work processes
and data collection to improve
productivity
> Proactively analysis and
reaction to potential problems
to increase efficiency and
better plan maintenance
> Increased equipment
availability and reliability
> Enabled field workforce to
improve decision making
By preventing a motor failure,
early detection led to
$70K Saved
Identifying a blocked plug on
a floating roof tank averted a
loss of
$120K
Estimated annual
maintenance savings of
$3-5Million$
Case Study: US refinery implemented a standards-based CMMS to better respond to the changing needs of the company
Results:
> Scalable asset management
to support company growth
> Familiar Microsoft®
Windows® work environment
> Improved management of
maintenance and operations
> Full accountability of work
orders, labor and materials
> Streamlined purchasing of
materials and services
Enabled the refinery to
manage its maintenance
activities and business more
efficiently than before
The software also enforces
data integrity, keeping all
data current and in check
The new software has been
widely accepted as a daily
utility and is used extensively
by employees
Closing the Loop
After installation of the new CMMS, the
refinery now has full accountability from the
moment that a person issues a work order,
requisitions materials, or contracts labor.
Asset management is a key application
used every day. They manage maintenance
activity and business more efficiently than
before. The software also enforces integrity,
keeping all data current and in check.
Case Study: Industrial gas company uses Predictive Analytics to detect compressor problems & avoid forced outages
Results:
> Detected equipment failures
before they occur
> Prevented unscheduled
system downtime
> Harnessed real-time
information that brings plant
operators and engineering
experts together to
collaborate
Predicting a
Compressorfailure
Using
Real-time Datato detect unforeseen events
Saving
+500K by preventing reactive
maintenance and unplanned
downtime
$
Closing the Loop
The company used a scheduled opportunity
outage to repair a damaged turbine engine.
Prior to a scheduled outage, the plant
notified the operations center of a vibration
sensor anomaly. As a result, the company
was able to use a planned outage to
investigate the compressor and found a
cracked impeller.
Case Study: US Refinery ‒ Equipment Performance Monitoring
Problem:
> Compressor performance degrades over time
> Operations doesn’t know impact of performance
degradation
> “How is the compressor operating currently?” vs.
“How it is supposed to operate?”
Solution:
> Model-based performance monitoring
> Track actual efficiency vs. design
> Track impact of cleaning on energy usage and
throughput
Results:
> Know when it is economical to service compressors
> Prevent damage to multi-million dollar machinery300Time (Days)0 100 200
DE
0
Design Eff(best)
Effi
cien
cy
Flow
Day 0
Day 200Day 100
Design
Case Study: Indian Refinery ‒ Crude Preheat Train Monitoring
Situation:
> Complex network of heat exchangers in crude preheat
train
Objective:
> Minimize crude furnace firing (i.e., fuel consumption
by maximizing preheat efficiency
Results:
> Increase in crude furnace inlet temperature
> Improvement in maintenance efficiency
> Identification of instruments needing recalibration
Published daily reports - Bad flow measurements
Benefit: Decreased energy usage by improving heat integration and
optimization of exchanger cleaning, resulting in annual
savings of more than $1 million/year
Thank you for your time and attention!