Novel Frameworks for the Smart Next-Generation Wafer Fab Presenter: Steve Blaine, P.E.
Novel Frameworks for the Smart Next-Generation
Wafer FabPresenter: Steve Blaine, P.E.
Internet of Things….What is it?
• The IoT is a future where computers know all
there is to know about the “things” around them.
• Using data gathered without action by people,
computers will monitor and even control these
“things”.
• We will be alerted to potential problems and be
able to take action before actual errors occur.
• With our increased knowledge, we will
automatically reduce waste and lower costs.
http://kevinjashton.com/2009/06/22/the-internet-of-things/
Internet of Things….What makes it work?
• Smart Ubiquitous Sensors: Inexpensive, easily connected sensors provide more, and more accurate, data.
• Mobile Devices: Provide the ability to monitor and interact with systems and things - located almost anywhere
• Cloud Computing: Providing computer services from multiple redundant offsite locations available on the Internet
• Interoperable Systems: Connectivity between systems enables a level of collaboration that was previously impractical.
• Analytics: Changing data into information to gain insight, improve decision-making and optimize operations
Internet of Things….how does it change factory automation?
Not so new:
• Smart Sensors are already here…and as low cost as possible…
• Mobile Devices is a common control system feature now…
• Cloud Computing is not so different from off site servers and storage…
Really new:
• Interoperable Systems: Connectivity between systems enabling a much higher level of collaboration than has been practical until today.
• Analytics: Applying analytics to this combined set of data to gain insight into operations, improve decision-making and optimize operations
Applications of Analytics
• Pattern recognition– Identification of previous occurrences, combining data
from multiple systems
– Algorithm: Time series data analysis (convolution, blind source separation)
• Anomaly detection– Identification of multivariate data excursions from norm
– Algorithm: Multi-dimensional clustering
• Predictive analysis– Tools that determine the probability of an outcome or
event by identifying cause and effect relationships
– Algorithm: Random forests (trained ‘forest’ of decision trees)
• Visualization– Easily comprehended “dashboards” to convey
information from a variety of sources
What are Analytics?
Analytics
• Exploratory data analysis (EDA), where new features in the data are discovered
• Confirmatory data analysis (CDA), where existing hypotheses are proven true or false
• Uses data visualization (e.g. dashboards) to communicate insight
Example – Predictive Analytics
Iron oxide peak from leaching ductile iron pipe
Early identification of aggressive water problem
saved the utility $20M in early replacement costs
Example – Analytics Visualization
http://www.skyfoundry.com/file/52/Case-Study-SkySpark-in-an-MV-Application.pdf
Example – Analytics Dashboard
http://blogs.msdn.com/cfs-filesystemfile.ashx/__key/communityserver-blogs-components-
weblogfiles/00-00-00-30-07-metablogapi/2251.image_5F00_658322A3.png
Application of Interoperability:
Complex Event Processing
• Makes decisions in real time with continuous streams of data
• Can provide advanced alarm management while suppressing false alarms
• Predicts impacts to scheduling (B2MML links to ERP/MES)
• Can change systems from “automatic” to “autonomous” –reacting appropriately even when faced with a situation never envisioned by the designer
What is Complex Event Processing?
Complex Event Processing
• Event Driven Architecture (EDA): software designed to detect, use, and react to events.
• Time-based, operates on continuous streams of data coming from many sources.
• Understands and manages stream relations. Assumes high event rate.
• Detects patterns in data.
• Produces output event streams or individual events.
Control Systems Today
Semiconductor Facility Management System
`
VFDs
Sensors/Devices
VFDs
Sensors/Devices
Ethernet
(OPC)
Remote IO
Networks
(Ethernet/IP, DDS)
Field Networks
(Profibus, Modbus,
HART…)
Systems Components
FMS System Equipment
• SCADA Workstations
• SCADA Servers
• Large PLCs/Controllers
• Small PLCs/Controllers
• Field Devices
• VFDs
FMS System Connections
• Ethernet
• Remote I/O Networks
• Field Networks
• Field Wiring
Ethernet
`
EXAMPLE: Semiconductor Facility Management System
60,000 ft (~5500 m2)
FMS Example
PLC/Field Equipment
• 15 Large PLCs/Controllers
• 15 Small PLCs/Controllers
• 100 VFDs (non-FFU)
• 2,500 Field Devices
SCADA Equipment
• 10 SCADA Workstations
• 4 SCADA Servers
• 50,000 SCADA points
2,500 “Hard” I/O Points50,000 “Soft” SCADA Points
Semiconductor Facilities – Control Systems Often in Separate “Silos”
•FMS
- Mechanical
- Cleanroom HVAC
- Process
- Vendor Systems
•Electrical
- Power Monitoring
- Generators
- UPS Status
- Auto Transfer Switches
- Substations
•Specialty Gas
- Gas Cabinets
- Pressure Sensors
- Flow Sensors
- EMO Buttons
•Specialty Chemical
- Pumps
- Tank Levels
- Liquid Leak Detectors
•Toxic Gas Detection
- Gas Detectors
- Horns/Strobes
- Tool Shutdown
•Fire Alarm
- Smoke Detectors
- Fire Dampers
- Horns/ Strobes
- Seismic Detectors
•BMS
- AHUs
- VAVs
- CAVs
- CO2 Sensors
Total = 240K Points
OPC BACnet Modbus Proprietary OPC Proprietary
•FFUs
- Fan Status/ Speed
- Zone Alarms
- Power
- EMO Buttons
Proprietary Proprietary
50K
Points
30K
Points50K
Points
10K
Points
20K
Points20K
Points
50K
Points
10K
Points
Reference Example from Data Centers –Data Center Integrated Management
•Mechanical/ HVAC
- Chillers Towers
- Air Handlers
- Exhaust Fans
- Terminal Units
•Electrical
- Power Monitoring
- Generators
- UPS
- Automatic Transfer Switches
- Substations
•Computer Servers
- Temperature
- HDD Usage
- CPU Usage
- Network Usage
Data Center Integrated Management
BACnet ModBUS SNMP
DCIM
Benefits
• Effective in providing a “single seat” monitoring capability
• Dashboards
• Capable of comparing operations at multiple data centers
Challenges
• Computer system centric (i.e. expertise favors one of the three systems – usually computer servers)
• Recreates capabilities already present in the other systems
• Very costly
Semiconductor Facilities –Control Systems Still in Separate “Silos” but with FMS “On Top”
•FMS
- Mechanical
- Cleanroom HVAC
- Process
- Vendor Systems
•Electrical
- Power Monitoring
- Generators
- UPS Status
- Auto Transfer Switches
- Substations
•Specialty Gas
- Gas Cabinets
- Pressure Sensors
- Flow Sensors
- EMO Buttons
•Specialty Chemical
- Pumps
- Tank Levels
- Liquid Leak Detectors
•Toxic Gas Detection
- Gas Detectors
- Horns/Strobes
- Tool Shutdown
•Fire Alarm
- Smoke Detectors
- Fire Dampers
- Horns/ Strobes
- Seismic Detectors
•BMS
- AHUs
- VAVs
- CAVs
- CO2 Sensors
OPC BACnet Modbus Proprietary OPC Proprietary
•FFUs
- Fan Status/ Speed
- Zone Alarms
- Power
- EMO Buttons
Proprietary Proprietary
50K
Points
30K
Points50K
Points
10K
Points
20K
Points20K
Points
50K
Points
10K
Points
Facility Management System
Total = 240K Points
M2M - Machine-to-Machine vs Many-to-Many
• “M2M” exists today – but usually as “Machine-to Machine”
communication
• “M2M” in the IoT is a wider concept - encompassing “Many-to-Many”
communication enabled by Big Data.
Big Data
M2M : Machine-to-Machine M2M : Many-to-Many
IoT = A System of Systems
The IoT is a “System of Systems” with both “horizontal” and “vertical” communication.
Source: http://electronicdesign.com/embedded/understanding-protocols-behind-internet-things
Flow Patterns –How will these systems interact?• Push
is a one-way communication between two parties in which a server sends data to a pre-defined client that receives the data.
• Request/Responseis a synchronous communication between two parties. A client sends a request to a server. The server receives the request and responds with the requested data back to the client. The client must wait for the response until it receives a response or until the request times out.
• Subscribe/Notifyallows asynchronous communication between two parties without the client waiting for the server response. The client indicates interest in a service by sending a “subscribe-call” to the server. The server adds the subscription to the address of other clients and sends notifications when data is ready to be sent.
• Publish/Subscribeallows a loose coupling between communication partners. Services offering information are advertised through a broker. Clients declare their interest in certain information by subscribing to the broker which makes sure the requested information flows between the service and the client. This pattern is most suitable for the IoT.
IoT Solution 1: Message BrokerPublish/Subscribe is fundamental to the IoT for two key advantages:
– Loose Coupling
– Scalability
The implementation of a Publish/Subscribe flow is typically
achieved by a middleware solution based on a Message Broker:
IoT Solution 2: Data Bus
Although Message Brokers will certainly be key elements in S2S and D2S
interaction, no single technology or protocol will monopolize the IoT scene,
and different technologies and protocols will be used for different types and
levels of interaction.
For instance the DDS protocol for D2D interaction implements a broker-less
Pub/Sub communication based on a Data Bus.
Proposed Architecture for Semiconductor Facilities
FMS BMSGases/
ChemicalsPower
MonitoringOther Systems
ComplexEvent
Processor
Analytics Engine
Web-based Query Tool
Feedback to change system operation
Data Bus / Message Broker
Selected Data Streams
NoSQL Database
Web-based Monitor
Computerized Maintenance Management
Production Systems(ERP)
Alarming & Paging System
Similar to Architecture used at STAR:
https://indico.cern.ch/event/258092/session/7/contribution/96/material/slides/0.pdf
Today’s Control Systems vs. IoT
Today’s Control Systems
• More IntegratedOrganized in a pyramidal way with a
single proprietary system on top
performing several common tasks.
• Tightly Coupled
(Client-Server)The client cannot post messages to
the server unless the server
process is running and available
• Automated SystemsSystems control according to
programmed functions without
human intervention
IoT
• More ModularSubdivided in more modules and skids,
each one performing it’s function and
sharing information with the others
• Loosely Coupled
(Publish-Subscribe)Publishers are loosely coupled to
subscribers, and need not even know
of their existence
• Autonomous SystemsSystems react appropriately even when
the system’s designers did not
anticipate the current system state.
Enabling Technology
Networking
• Wireless
• Mobile Devices
• Converged IT/AT
• IP v6
Devices
• MEMS
• RFID Tags
• Energy
Harvesting
• Geo-location
• SOC Security
All of these things are driving the creation of the IoT…
but more is needed….
Computing
•Cloud Services
•Security
What Else is Needed to Make It Work?
Middleware
• Message Bus / Brokers
Goal: Allow messaging between
applications, for instance
Publish/Subscribe messaging– Example: Apache ActiveMQ
• Enterprise Server Bus
Goal: Implement communication
between mutually interacting software
applications in a SOA– Examples: IBM WebSphere, JBoss, Oracle ESB, WSO2
Concepts
• SOA (Service Oriented Architecture)
• CEP (Complex Event Processing)
• EDA (Event Driven Architecture)
Messaging Protocols
•MQTT
(Message Queue Telemetry Transport)– Simple, lightweight (D2S, S2S)
•AMQP
(Advanced Message Queuing Protocol)– More feature-rich and complex (S2S)
•DDS (Data Distribution Service)– Oriented to real-time applications (D2D)
•OPC UA (OPC Unified Architecture)– Oriented to industrial applications (D2S, D2D,S2S)
Existing Standards and Technologies
Data
• XML
• JSON
• SQL
• NoSQL
Presentation
• HTML 5.0
• CSS
• Javascript
IoT will use many existing standards and technologies.
For instance:
Communication
•OPC/UA
•DDS
•MQTT
•AMQP
Web Services
•SOAP
•REST
Some of these are different from the automation
tools we use today…
Key Challenges: Semantics
Source: http://www.iot-a.eu/public/public-documents/d1.5/view
What Should You Do Today?
Invest in Controls and Networking• IoT is not here yet but the “Intranet” of things exists now.
Invest in Truly Open Systems• Ensure that the data you need is exposed and available
• Avoid the “who’s on top” dilemma
Invest in Information• Information in your systems is valuable
• Information about your systems is also valuable
Think Big! • What would you do if you knew everything about everything?
One last thought: Build it Yourself!
• Vendors can provide the tools – but no one can build an IoT
application without detailed knowledge of your systems
• Become self sufficient or hire a system integrator
Thank you!
Special thanks also to:
AMQP Advanced Message Queuing Protocol = S2S messaging protocol feature
B2MML Business To Manufacturing Markup Language = an XML implementation of the ANSI/ISA-95 family of standards
BACnet Building Automation and Control Network = communication protocol for HVAC and associated equipment
BMS Building Management System = control system for mechanical and electrical HVAC equipment
CDA Confirmatory Data Analysis = type of analytics where existing hypotheses are proven true or false
CEP Complex Event Processing = method of tracking and analyzing data and deriving a conclusion
CPU Central Processing Unit = core computer element that carries out instructions
CSS Cascading Style Sheets = language used for describing look and feel of web pages, separates document content from presentation
D2D Device to Device = communicaiton between devices
D2S Device to Server = communication between devices and servers
DCIM Data Center Integrated Management = system to combine computer, facility and electrical management for data centers
DDS Data Distribution Service = messaging protocol used fro D2D communication for real-time applications
EDA Exploratory Data Analysis = type of analytics for manual querying to dscover new features
ERP Enterprise Resource Planning = business management software for manufacturing
FFU Fan Filter Units
FMS Facility Management System
HDD Hard Disk Drive
HVAC Heating / Ventilation / Air Conditioning
IoT Internet of Things
IP v6 Internet Protocol Version 6
JSON JavaScript Object Notation = open standard format using human-readable text to transmit data objects consisting of attribute–value pairs
M2M Machine to Machine
MES Manufacturing Execution System
ModBUS Serial communication protocol connecting industrial devices
MQTT Message Queue Telemetry Transport = Lightweight D2S or S2S Messaging protocol
NoSQL Non-relational Structure Query Language = mechanism for storage and retrieval of data not using tables as done in relational databases
OPC Open Platform Communication = standard for communication of data between control devices from different manufacturers
OPC UA OPC Unified Architecture = successor to OPC providing cross platform service oriented architecture
PLC Programmable Logic Controller
REST Representational State Transfer = software architecture for building scalable web services
S2S Server to Server = communication between servers
SCADA Supervisory Control and Data Acquisition
SOA Service Oriented Architecture
SOAP Simple Object Access Protocol = protocol for exchanging information via web services
SOC System on a Chip = integration of all computer components on a chip
SQL Structured Query Language = programming language designed for managing data in relational databases
UPS Uninterruptible Power Supply
VFD Variable Frequency Drive
XML Extensible Markup Language = set of rules for encoding documents in human-readable and machine-readable format
Glossary