IIoT – Practical Implementation · IIoT –Practical Implementation 9/2/2016 Jim Kosmala – Vice...
Transcript of IIoT – Practical Implementation · IIoT –Practical Implementation 9/2/2016 Jim Kosmala – Vice...
IIoT – Practical Implementation9/2/2016
Jim Kosmala – Vice President Engineering, Okuma
Andy Henderson, PhD – Industry Analyst, GE Digital
Industrial Internet of Things
• Why Get Connected
• 3 Reasons some are not connected yet
• Applications in Production
Why Get Connected ?
MachineIntelligent
TechnologyMT Connect
GE Brilliant Factory
= Opportunity
5%
Connected
Lack of Machine
Knowledge
Avg. 65% Asset
Utilization - time
producing parts
Inefficiency +
1. Get Connected
2. Get Insights
3. Get Optimized
Machine Health/Predictive-Preventative Maintenance
Scalable to Fit Your Needs
Software
Hardware
Data Communication
Machine Health
Predictive / Preventative Maint.
Quality, SPC
Energy Management
Communication
Scalable
Small
Medium
Large
Video Communication
MTConnect
OSP Control
MTConnect
Non-OSP Control
Predix - Embedded Predix - Ready
New Collaboration
Predix Machine
(Run-time app)
Predix Machine
(Run-time app)
Predix Cloud
Predix Client
Apps
Predix Field
Agent box
PC Data Server
Manufacturing at GE
POWER
~$30B
66K EMPLOYEES
OIL & GAS
$18.7B
46K EMPLOYEES
AVIATION
$24B
45K EMPLOYEES
ENERGY
MANAGEMENT
~$11B
47K EMPLOYEES
RENEWABLE
ENERGY
~$9B
13K EMPLOYEES
APPLIANCES
& LIGHTING
$8.4B
25K EMPLOYEES
TRANSPORTATION
$5.7B
13K EMPLOYEES
HEALTHCARE
$18.3B
55K EMPLOYEES
2014 REVENUES
• $50B Manufacturing Spend
• 400+ Plants
• 1% Efficiency > $500M Cost Savings
Manufacturing improvements are very
important to GE!
Industrial Businesses:
New Opportunities
SalesNetwork
GE’s Brilliant Factory
15
• 1.5M square feet of Manufacturing Space
• Machining, Welding, Assembly, Spray Coating
• New make & Repair
• Components and Full Assembly
Greenville: Gas Turbines
16
7HAPower Output:
7HA.01 = 275 MW
7HA.02 = 337 MW
Power Output: 198 MW
7F.04
7E.03Power Output: 91 MW
9F.03
Power Output: 265 MW
Power Output: 231 MW
7F.05
ElectricProcess GassesBldg Mgmt
Compressed Air
Durable GoodsTool Life Mgmt
Process DataUtilization
PM TimersShop Floor Dashboard
Predictive
Consumable
Management
Energy
ManagementMachine
Health
Process
Optimization
Greenville: Gas Turbines
• 1000hp motor for #2
main air compressor
• Stopped before major
catastrophic failure
-------------------------------------------------------------------------------------------------------------
Cost Avoidance:
Prevented catastrophic failure
Eliminated need for new motor
Saved ~$###k in overhaul costs
Big Data Info Action $$$
Brilliant Factory Win – Machine Health
-------------------------------------------------------------------------------------------------------------
Cost Avoidance:
• Prevented unplanned system
downtime
• Avoided $##K/year in wasted gas
• High Argon flow during plant
shutdown led to “Treasure Hunt”
• Leak found at Coatings Furnace
• Leak totaled approximately 200cfh
Before find it, fix it
After find it, fix it
Big Data Info Action $$$
Brilliant Factory Win – Energy Management
Big Data Info Action $$$
-------------------------------------------------------------------------------------------------------------
Brilliant Factory Tools
Shift-By-Shift Utilization
& Completed Operations
Live Feed of
Machine Activity
Defect & Work-
Order Tracker
Machine Utilization – July 36.04%
Machine Utilization -
November50.9%
Cost Avoidance over 12
months$##k
Machin
e U
tiliz
ation %
Problem:
• Mill: Low utilization
• High demand for part
Issues Identified:
• Machine not completing
full cycles
• Defects and re-cuts
• Machine idle time
between parts
Improvements Made:
Improved cutting tool
utilization
Fixture and part loading
enhancements
Standard work schedule
for operators
Brilliant Factory Win – Process Optimization
More Sensor Analysis
Unsupervised Learning / ClusteringAutomatic Identification of Anomalies
Machine Learning (Spindle Load - Okuma MB 5000H)
1. Thermal Sensors Feedback To CNC
2. Vibration Sensors
Feedback to CNC
4. CNC Compensates
Axes Independently
3. CNC Calculates Thermal Error
Estimate thermal
deformation
Temperature &
Operation data
Thermal deformation
compensation… and other sensor data