smart manufacturing journey alcoa smelting and...
TRANSCRIPT
© Copyright 2015 OSIsoft, LLC
Presented by
SMART Manufacturing Journey Alcoa Smelting and Casting
Bruno Longchamps
Pierre Boutin
SMART Manufacturing Journey Alcoa, Global Smelting and Casting
2
Bruno Longchamps, P.Eng SMART Manufacturing, Program Manager
OSIsoft User’s Conference, San Francisco, April 2015
Presentation content
Alcoa SMART Manufacturing Journey Overview
Users Engagement
Anode Tracking
Cast Data Collection & Integration
Conclusion
Questions
3
Alcoa Inc.
4
A global leader in lightweight metals
technology, engineering and manufacturing
Revenues of US$23.9 billion in 2014
59,000 employees in 30 countries
At December 31, 2014
Aluminium Production Process
5
Electrode:
Anode
production
Potroom:
Liquid aluminum
production
Casthouse
Salable solid aluminium
production
c
o
k
e
P
i
t
c
h
6
Smelting Operation
Rodded Anode
Potroom
Casthouse Furnace
Finished Products
Aluminum production requires very large manufacturing sites
With time, the process control and manufacturing systems have multiplied and become
heterogeneous
Each production system was an isolated data island with limited or no data history
Common operation best practices and new technology deployment were very difficult to implement
7
Why Alcoa invested in the SMART Manufacturing Initiative ?
SMART Manufacturing Program
• Well integrated, fine
granularity, long term
production data available
• Increase collaboration and
sharing at the plants and
between locations
• Achieve significant
savings
Solution Results / benefits Challenges
• Aluminium market has
been financially challenged
for several years
• Improve competitiveness
• Enhance operational
excellence
• High retirement rates, loss
of SMEs
• Establish the SMART
Manufacturing program
• OSIsoft enterprise agreement
• Deploy a robust infrastructure
« cookie cutter » project in all
plants
• Mobilize key employees
SMART Manufacturing is about
integration of data with process expertise
to enable proactive and intelligent
manufacturing decisions in dynamic
environments
8
SMART Manufacturing – Main improvement axes
Data
Maturity Model
Before:
After :
9
SMART Manufacturing – Real plant wide data infrastructure
10
Note for Lista: QLC not installed and the initial data collection phase is not completed
Site Name Total
GPM Global BU 1,559,207
GPM Warrick Site 1,246,100
GPM Alumar Site 1,193,900
GPM ABI Site 1,042,900
GPM Baie-Comeau Site 951,450
GPM Portland Site 675,770
GPM Mosjoen Works Site 628,190
GPM Fjardaal Site 554,000
GPM Deschambault Site 501,460
GPM Massena West Site 337,590
GPM Lake Charles Site 30,751
Warrick Power Plant 22,358
GPM Lista Site 6,548
Grand Total 8,750,224
The journey to the implementation of the SMART Mfg program
2010 Program Definition (OSIsoft support)
2011 Proof of concept (Deschambault plant, QC)
2011(Q4) Entered the OSIsoft Enterprise Program
2012-2014 Aggressive Global Deployment (13 plants)
2014 EA extension to other plants
2015 Additional deployments (2 to 5 plants)
11
Approx. one deployment start-up every 2 to 3 months
Each deployment required a 6 to 9 months effort
Global BU instance added in 2014
12
SMART Mfg. – Deployment Schedule
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4
Deschambault
Massena West
Baie Comeau
Mosjoen
Fjardaal
Lake Charles
Warrick
Portland
ABI Schedule at Risk
Reybec Behind Schedule
Global BU Instance
Sao Luis On Hold
Lista Closed
Intalco
San Ciprian
Ma'aden
2011Site
On Schedule
2015
Project Found
and Startup
Value Realization
ACES
2012 2013 2014
Deployment &
Data collectIon
SMART Manufacturing – Geographical localisation
13 All Data is random, fictional data to show functionality only
Process Equipements
PI Data Archive
MES/LIMS
MS SQL BI
ERP
(Oracle)
AF
& E
F M
odel
& B
I Cubes
Microsoft Sharepoint PI
WebParts
MS SP
Web Parts
SSRS/Power
View Report
Microsoft Tools Excel, Power Pivot, BI Tools...
OSIsoft Tools PI ProcessBook, PI Coresight, ...
Data Collection and Analytic Data Visualisation
SMART Manufacturing – Functional model
14
SMART Manufacturing – Implantation diagram
xxx-SMART(DNS alias)
xxx-MES-SQL-CL1Windows ClusterTITLE
SMART Manufacturing System Architecture
March 12th, 2014
xxx-SMT-PI(PI Collective)
xxx-SMT-AF (NLB)
xxx-SMT-SQL-CL1(Windows Cluster)
CMI - Control System LAN (one or more)
LEGEND
· mPI = OSIsoft Enterprise Agreement (EA)
· P = Primary Machine
· S = Secondary (Redundancy)
· HA = High Availability
CMI Prod Domain - DMZ
CMI - Plant Manufacturing LAN
Office / Business LAN – (Business Domain)
PI ClientsPI ProcessBook
PI Datalink
IE (WebParts & Coresight)
Analysis Tools & Displays of Manufacturing Related Data
Active
Directory
Trusting
Domain
Server
CMI DFS ServerCMI DFS
Mapped drive M:\
DFS Files Share
SMART
ManagementmPI
Excel
PI-SMT
PI Datalink
PI OLEDB Ent
Bomgar Station (support)
Admin & Maintenance Apps
xxx-SMT-MGM-01
PI ServermPI
PI Server [HA]
PI AF [HA]
PS
PI AF ServicesmPI
Pi Notification [HA]
PI ACE [HA]
PI AF Analytics (AF-01)
Calculation EngineNotifications
SMART ServicesmPI
SMART Calc engine
Site coded calculation
PI-Agent proxy for PCS LAN PI-Nodes
Real-Time Calculations PI-Agent proxy for PCS LAN
SMART SharePoint
Entreprise Server
mPI
PI WebParts
PI Datalink Server
PI OLEDB Enterprise
Real-Time & Historical Plant Floor Data
xxx-SMT-WSS-01
xxx-APM-DFS
MS-SQL SMART mPI
SQL Server 2012 Enterprise
High availibilty
Clustered
SSAS / SSIS / SSRS
AF , EF, SharePoint DBPlant MI
Datawarehouse
PS
Historian ServerPlant assets and
hierarchy
PS
xxx-SMT-PI-01
xxx-SMT-PI-02
xxx-SMT-SVC-01xxx-SMT-SQL-01
xxx-SMT-SQL-02
APMSDG Domain
QLCL&K
CMIFireWall
Site
MS-SQL MES
ServerSQL Server 2012 Std
High availibiityMES Apps and Services
xxx-SMT-MES-01
xxx-SMT-MES-02
xxx-SMT-DMZ-TS(NLB)
Client Tools TSmPI
Excel
PI ProcessBook
PI Datalink
PI AF Client
PI-SMT
SMART & MES
xxx-SMT-TS-4A
xxx-SMT-TS-4B
xxx-SMT-PCN-TS(NLB)
Terminal ServerPI-SMT
PI-Datalink
SMART & MES Apps
xxx-SMT-TS-3A
xxx-SMT-TS-3B
xxx-SMT-WEB(DNS Alias)
SMART WEBmPI
IIS SMART App. Web Service
Pi Web Services
PI Coresight
SMART DashboardAnd WEB apps.
xxx-SMT-WEB-01xxx-SMT-AF-01
xxx-SMT-AF-02
xxx-SMT-NOD-01
NodemPi
SMART NODE(VM Guest)
PS
xxx-QLC3-LDR-A
xxx-QLC3-LDR-B
mPI
Proxy: xxx-SMT-SVC-01
15
User Engagement
Pierre Boutin, P.Eng, MPM Manager, Global Manufacturing Solutions Delivery
16
OSIsoft User’s Conference, San Francisco, April 2015
Organization structure for change management
17
Sector
leaders
Plant Global
Regional leaders Global
SMART technical team Global
Technical lead Value lead
OSIsoft team
Plant sector lead/area super-users
End-users
Global
End-users
Key success factors
18
Data visibility
Key success factors
19
Key success factors
20
Post installation follow-ups
Train users
Answer questions
Fix issues
Share best practices from other locations
Escalate
Detect lack of engagement before it is too late
Report value to sponsors
Keep communication channels open…
Working across silos at the plant level
21
User engagement at the plant level
Electrode Potline Casthouse Environ. Energy
SMART mfg horizontal integration
Working across silos globally
22
Anode tracking Pierre Boutin, P.Eng, MPM Manager, Global Manufacturing Solutions Delivery
23
OSIsoft User’s Conference, San Francisco, April 2015
Anode tracking R&D project
• Holistic insight over anode life
• Anode performance linked to
raw materials and anode
fabrication data
• Development of anode
manufacturing best practices
• Improved Raw Material
selection process
Solution Results / benefits Challenges Technical
• Physical anode identification
• Linking anode data available at
different point in time
• Continuous and discrete
process involved
Business
• Cost of raw materials
• Establishing anode fabrication
best practices
• Read anode numbers with OCR
camera
• Store sampled data in PI Server
• Use Event Frames to organize data
in time
• Integrate all data sources with the
MS SSAS toolset (BI cube)
The goal of the anode tracking project is to
make anode genealogy data (7 to 8 weeks
from start to finish) easily accessible to support
process improvements.
24
Work breakdown structure
25
Physical Tracking
Data Processing
Process optimization
Bake Rodding
Electrolysis Paste
Anode tracking
Database
PI
System • Process modelization
• Reaction plan
• Preventive/proactive actions
• Predictions
• Process optimization
• Best practices development
• Data collection
• Structure/organize data
• Manufacturing Intelligence
Data – Continuous process
26
Paste plant
Rodding
Automated reading
Electrolysis
Butt treatment
Anode numbers
Rod numbers
Rod numbers
Crane
s
Rod numbers
Anode positioning
Cranes
Rod numbers
Rodded anodes
Butts
Anode numbers
Anode weighing
Data – Discrete process
27
Challenges
28
Data processing
29
Asset Framework :
equipment hierarchy
PI Data
Server
Microsoft Excel
Lab,
MES,
other
sources
Event Frames :
organize data in time
Automation
Raw data
Data extraction / exploitation
Microsoft SQL-Server
Analysis Services (SSAS)
dialog
30 All data is random, fictional data to show functionality only
Final result
31
Data sampled at different points in time are all linked to the same anode
All data is random, fictional data to show functionality only
Cast data collection and integration
Bruno Longchamps, P.Eng SMART Manufacturing, Program Manager
32
OSIsoft User’s Conference, San Francisco, April 2015
Cast data collection & integration
• Well integrated, right
granularity cast data
• Increased monitoring,
troubleshooting and
analytical capabilities
• Reduced scrap and
improved product quality
Solution Results / benefits Challenges
• Huge volume of data
• Adaptive and generic data
capture
• User driven data capture
enhancements
• Time series and relational
data integration in one spot
• Use Event Frames to capture
and organize data in time
• Extend EF internal data
capability
• Integrate all data sources with
the MS SSAS toolset (BI cube)
• Automated EF structure and
data transfer to the cube
The goal of this initiative was to deliver an
efficient and flexible cast equipment's data
collection engine and to combine resulting
data with other data sources such as
elaboration, quality, laboratory and tracking
33
Casting Operation Overview
34
Casting operation:
When metal preparation
completed in the furnace
The furnace is feeding the
liquid metal to a casting table
Casting table is forming and
cooling the metal
The Casting Pit elevator is
going down from the top
position to the bottom (Cast
length)
1 to 3 hours per cast
Metal
Furnace
Up to 30
Feet Long
Elevator
Casting Data Collection & Integration technical challenges
Summarize the volume of time series data: 50 sensors and set
points x 3 hours cast x 1 sec frequency = approx. ) 0.5 M values per
cast;
Ability to capture cast data for a specific internal process step like:
start-up time, steady state time, stoppage time;
Ability to capture cast data at a specific time (on critical event);
Deliver a complete process data integration (from all sources);
Be able to quickly compare, slice and dice from various angle critical
cast data;
Finally, from the cast summary data, be easily able to go back to the
detailed data in one mouse click;
35
Solution : Casthouse Data Integration Concept
36
AF Data Model
PI Data
Server
Microsoft Excel
Other
sources
Event Frames
Automation
MES &
LIMS
Casthouse
Casthouse AF Data Model
37
Casting Operation Event Frames Boundaries
38
Cast Start =
Event Frames
(EF) Start
Cast Stop =
Event Frames
(EF) End
Cast Length
Capture data for specific process phase (time window)
39
Step 5 to 7 = Startup (ramp-up)
Step 8 = Steady phase
Cast Start Cast End
Step 5 to 7 = Startup (ramp-up) Step 5 to 7 = Start-up (ramp-up)
Step 9 to 10 = Stoppage
Capture Data during the Steady State phase only
40
Based on predefined Cast Steps (Steady phase step = 8)
Data is summarized for the
cast steady state time
window as configured: Step Start @ mm
Step Stop @ mm
Step: 2:49:59 hours
41
Cast Length >=
7,500 mm
At 3:28:54 AM
Additional Data capture capability: Data at a specific moment
Capture information at a specific moment (Like a picture)
42
Cast Length >=
7,500 mm
At 3:28:54 AM
Can always go back to the detail with imbedded PI Coresight hyperlink
43
BI Cube needed for the Integration of all data sources (EF and MES)
44
• Summary of all Cast critical data in one place • Can look at a large volume of data quickly • Can do cast to cast comparisons • Can go back to the detailed data in one click (PI Coresight hyper link)
44
Can always drill down to the detail with PI Coresight from the CUBE
45
Cast display in PI Coresight – Focus on Start and End
46
Event Frames – Furnace Preparation EF (if time allows)
47
Metal Temperature at the
silicon addition time
Important EF attribute can be displayed in a KPI (if time allows)
48
Note: Numbers displayed on this slide do not represent real operation data
780
656
700
600
© Copyright 2015 OSIsoft, LLC 49
Contact information
Bruno Longchamps, P. Eng.
SMART Manufacturing, Program Manager
Alcoa Global Primary Products
Pierre Boutin, P. Eng., MPM
Manager, Global Manufacturing Solutions Delivery
Alcoa Global Primary Products
© Copyright 2015 OSIsoft, LLC
Questions
50
Please wait for the microphone before asking your questions State your name & company
© Copyright 2015 OSIsoft, LLC