Transcript Contoured 3D Printing of Fiber Reinforced polymer

Latency and Communication Challenges
in Automated Manufacturing
Kira Barton
Department of Mechanical Engineering
University of Michigan
November 3rd, 2016
Data flow in automated manufacturing
Current Manufacturing Automation
Digital Manufacturing – new enabling
technologies:
•
•
•
•
•
•
Cloud computing
Internet of things
MTConnect
Virtualization
Service-oriented architectures
Advanced computing technologies
Current Manufacturing Automation
Increased requirements on
latency and communication
Challenges with Real-time Data - GM
Challenges with Real-time Data - GM
Sensing
Real-time
Data
Machines
Inventory
People
What to collect ?
Smart Sensing?
Challenges with Real-time Data - GM
Sensing
Real-time
Data
Plant
Monitoring &
Control
Machines
Main Server
Internet
Inventory
People
What to collect ?
Smart Sensing?
What/When/Where
to transfer?
How to provide
feedback?
Challenges with Real-time Data - GM
Sensing
Real-time
Data
Plant
Monitoring &
Control
Remote Monitoring
& Control
Machines
Analysis
Main Server
Internet
Real-time
Data
Inventory
People
What/When/Where
to transfer?
What to monitor?
How to analyze?
What to collect?
Smart Sensing?
How to provide
feedback?
How to control?
Challenges with Real-time Data - GM
Sensing
Real-time
Data
Plant
Monitoring &
Control
Remote Monitoring
& Control
Product & Mfg.
Engineering
Real-time Data
Machines
Analysis
Main Server
Internet
Real-time
Data
DESIGN
Maintenance
Inventory
People
PLANNING
What/When/Where
to transfer?
What to monitor?
Anomaly
detection
Learning
How to analyze?
What to collect?
Smart Sensing?
How to provide
feedback?
How to control?
What to do with the
information?
Decision making?
Challenges with Real-time Data - GM
Sensing
Real-time
Data
Plant
Monitoring &
Control
Remote Monitoring
& Control
Product & Mfg.
Engineering
Real-time Data
Machines
Analysis
Main Server
Internet
Real-time
Data
DESIGN
Maintenance
Inventory
People
PLANNING
What/When/Where
to transfer?
What to monitor?
Anomaly
detection
Learning
How to analyze?
What to collect?
Smart Sensing?
How to provide
feedback?
How to control?
What to Communicate and how?
What to do with the
information?
Decision making?
Network requirements to be
addressed
Control: high fidelity for decision making
Diagnostics: data storage, analytics, learning
Safety: fast and accurate response times
1. Which sensors to include at each level – Machine, cell,
system, factory
2. How to convert and transfer data between enterprise levels?
3. What are communication needs and latency impacts?
Network Partitioning for Control,
Diagnostics and Safety on Ethernet
Ethernet I/O functionality domains
• Control: Medium data volumes, high determinism
• Diagnostics: High data volumes, high speeds, low
determinism
• Safety: Small data volumes, very high determinism
Network Partitioning for Control,
Diagnostics and Safety on Ethernet
Secure Cloud
Infrastructure
Central Switch
System Switch
Control
Control
Control level
Control
Control
Sensing
Machine level
Firewall
System Switch
Control
Network Partitioning for Control,
Diagnostics and Safety on Ethernet
Secure Cloud
Infrastructure
Central Switch
System Switch
Control
Control
Control level
Control
Control
Sensing
Machine level
Firewall
System Switch
Control
• Timing mismatch
• Network congestion
• Poor network
provisioning
• Noisy signals
• Data loss
Better
Performance:
Tracking Error
Worse
Network End-to-End Performance Cost
Impact of
sampling
Out of
Control
Ideal
Operating
Region
Networked
Control
Digital
Control
Network
saturation
inducing longer
delays
Unacceptable
Acceptable
A
B
C
Continuous
Control
PA
PB
Larger
PC
Smaller
Sampling Time
Impact of
sampling &
additional delays
Issues driving network performance
Can I use TCP/IP versus
EtherNet/IP or EtherCat?
Should I partition my
networks at different
levels?
Where is the delay
and delay variability
occurring?
What are the industry
defacto standards?
Should I put safety, control
and diagnostics on one,
two or three networks?
What is the tradeoff
cost of a decision?
What is the tradeoff
complexity of a decision?
What is the performance cost
of security or application level
protocols ?
Current methods to address issues
• Time synchronization algorithms
– close to the production tasks
– strict requirements on timing
– need context to the data to align timing
• Event-based control
– mitigate logic errors
– may result in time delays
• Multiplex network scheduling
– utilize alternative network interfaces for communication
– transmit over path with smallest estimated delay
– Improve software to address delays and communication issues
Automated manufacturing network
requirements
Data transfer
rate
Data transfer
load
Impact of
delays
Multi-plex
capabilities
Low-level
Control (CNC,
robots, etc.)
1ms or faster
Medium
Critical
(performance)
Typically no
High-level
control (PLCs)
~10ms
Low
Critical
(performance)
Typically no
Diagnostics
Slow
High
Minimal
Yes
Safety
Fast
Low
Critical
Potentially
Thank you for your attention!
Questions?