The Five Skills All Automation
Engineers Should Master
Standards
Certification
Education & Training
Publishing
Conferences & Exhibits
Your Presenter
Scott Sommer, PE, CAP
BS/MS Chemical Engineering
Automation for 35+ years
Married, 5 Children, 1 Grandchild & 2 Dogs
Lives in Phoenixville, PA (Valley Forge)
Hobbies:
Home Brewing
Automation Education & Training
ISA Trainer
CAP Steering Committee
Concept #1
Understand the Science Behind the
4-20ma Current Loop
Series Current Loop
Current to Voltage Conversion
Analog Transmitter Circuit Example
From EDN Network, February 18, 2014
Multiple Loop Devices
Smart Field Devices
HART Communication
• Uses Bell 202 modem to communicate digital signal
• 24VAC signal superimposed over 24VDC signal
• Frequency Key Shifting (FSK)
• Uses same wiring media as analog current loop (single shielded twisted
pair)
2200 Hz = 0
• 1200bps – “slow”, but powerful! Signal on Wire
1200 Hz = 1
0 0 1 1 1 1 0 0
Concept #2
Know How to Calibrate
An Analog Transmitter
DP Transmitter
Zero Error
100%
MEASURED
VALUE
ERROR CAUSED
BY INCORRECT ZERO
100%
0
TRUE VALUE
Span Error
100%
HIGH
SPAN
IDEAL
SPAN
MEASURED ERROR
LOW CAUSED
VALUE SPAN BY SPAN
100%
0 TRUE
VALUE
Linearity Errors
100%
MEASURED
VALUE ERROR CAUSED
BY NONLINEARITY
100%
0 TRUE VALUE
Adjusting for Errors (Calibration)
Calibration is an adjustment or set of adjustments performed on an
instrument in order to make the instrument function as accurately (error
free) as possible.
Verification/Calibration Block Diagram
INPUT
MEASUREMENT
(STANDARD)
PROCESS OR OUTPUT
INSTRUMENT
SIMULATED MEASUREMENT
UNDER TEST
INPUT (STANDARD)
POWER SOURCE
DP Transmitter Verification
DIFFERENTIAL mA READOUT
comm PRESSURE ± 0.015% ACCURACY
TRANSMITTER OR BETTER SERIES RESISTANCE
_ REQUIRED WITH
c
F + SOME TRANSMITTERS
+ o
I + mA
E _ 4-20 mA
+ m L
PROCESS
- m D
SIMULATOR
INCHES H2 O 0” - 250” H2 O
H L _ +
>0. 015% ACCURACY
24VDC
OR BETTER
POWER SUPPLY
VENT TO
ATMOSPHERE
>20 PSIG
AIR SUPPLY 120 VAC
Five-Point Calibration Check
• A credible calibration should include five points of reference:
– 0%
– 25%
– 50%
– 75%
– 100%
• Should include upscale and downscale checks to detect any hysteresis error
or lack of repeatability
Calibration Example (Chart)
INPUT VALUES OUTPUT VALUES
% of % Deviation
Actual Input Ideal As Found Values Calibration 1
Span 0 to 150” H2O (Error)
Values
(Span = 150” )
0% 0” 4 mA 4 mA 4 mA 0% 0% 4 mA 4 mA
25% 37.5” 8 mA 8.01 mA 8.02 mA .0625% .125% 8 mA 8.01 mA
50% 75” 12 mA 12.02 mA 12.03 mA .125% .1875% 12 mA 12.02 mA
75% 112.5” 16 mA 16.03 mA 16.04 mA .1875% .25% 16 mA 16.01 mA
100% 150” 20 mA 20.04 mA 20.04 mA .25% .25% 20 mA 20 mA
Calibration Example (Graph 1)
+ .6
+ .5
% DEVIATION FROM IDEAL OUTPUT
+ .4
+ .3
+ .2
+ .1
0
- .1
- .2
- .3
- .4
- .5
- .6
0 25 50 75 100
INPUT AS % OF SPAN
Calibration Example (Graph 2)
+ .6
+ .5
% DEVIATION FROM IDEAL OUTPUT
+ .4
+ .3
+ .2
+ .1
0
- .1
- .2
- .3
- .4
- .5
- .6
0 25 50 75 100
INPUT AS % OF SPAN
Calibration Adjustments
PROCESS ZERO POT
PRESSURE
SIGNAL OUTPUT 4-20 mA
SENSOR
CONDITIONING CONDITIONING OUTPUT
ANALOG TRANSMITTER SPAN POT
PROCESS
PRESSURE MEMORY
4-20 mA
SENSOR A/D MICROPROCESSOR D/A
OUTPUT
SMART TRANSMITTER DIGITAL
COMMUNICATIONS
Characterization and Digital Trim
CHARACTERIZATION 4 - 20 mA TRIM
PROCESS
PRESSURE MEMORY
4-20 mA
SENSOR A/D MICROPROCESSOR D/A
OUTPUT
DIGITAL
COMMUNICATIONS
SENSOR TRIM
Concept #3
Know How to Tune Simple and
Complex Control Loops
The Process Control System (Feedback Loop)
Typical Process Control Loop
Process Dynamics
• Dynamic response is a measure of the systems reaction, based on a function
of time
• Reactions include; a set point change, or a change of valve position or
responding to a process disturbance
DISTURBANCE
SP / MANUAL CHANGE OUTPUT RESPONSE
PROCESS
DYNAMICS
DEAD TIME +
FIRST ORDER LAG
First Order Lag
• Most process require some time to respond to changes
• A standard method of describing this time is the time constant
• A time constant is the time required for the output to change 63.2% in one time
period. The response is complete in 5 time constants
• The time constant is indicated by the symbol
Output (%) = 1s = 2s
100%
63.2 %
40%
20%
0% t
1 2 3 4 5 6 7 8 9 10
Time Periods (minutes or seconds)
Dead Time
STEAM
TIC
1
500 FEET 3" PIPE
PROCESS
FLUID
FLOW = 100 GPM
OUTLET TEMP
DISTURBANCE
Td TIME
Process Reaction Curve - FOPDT
Determining Process Parameters for a True
First Order Process with Dead Time (FOPDT)
The Ideal PID Block Diagram
1
Ti ∫
+ b
SP + Σ Kc Σ m
+ +
-
d
TD
PV dt
Block Diagram Representation of an Ideal PID Controller
Conventional Controller Modes
Symbol Description Mathematics Expression
One Mode
P Proportional m = Kc e + b
1
I Integral (reset) m = T e dt
i
Two Mode
1
PI Proportional + integral m = KC [e + T e dt]
i
de
PD Proportional + derivative m = KC [ e + TD dt ]
Three Mode
Proportional + integral 1 de
PID + derivative m = KC [e + T e dt + TD dt ]
i
Ziegler-Nichols Formulas for Open
Loop Tuning
CONTROLER GAIN RESET DERIV
MODES Kc Ti-Mins / Rep Td-Minutes
Prop. Only --------------- ----------------
KP q
0.9
Prop. + Reset 3.33 q ---------------
KP q
1.2
P+I+D
2.0 q 0.5 q
KP q
NOTE: KP = Process Gain q = Dead Time = Time constant
PID Analog Electronic Controller
SET ERROR
POINT
P
W
V
250 I
PROCESS I
VARIABLE I
P
FY
D
FV
FT FE
Possible “Good” Responses
Quarter-wave Decay
• A commonly used goal for “good response” to a step change upset is a ¼
wave decay or ¼ amplitude dampening
– Each overshoot is ¼ the previous
– Gives minimum error integration with stability
A B
B 1
A 4
• This is a good compromise between fast response and no overshoot
Tuning Maps for Gain and Integral
No Fast Integral
Integral
High Gain
• Gain is increased to decrease offset • Integral is increased to eliminate offset
• Too much gain can cause instability • Faster time integral can cause instability
Feedforward Control of Heat Exchanger
with FB
LIQUID IN
TT
TS P Wc P ( TS P – Ti ) Ti
FC
F =
H
FT
W S TEAM
+
F
-
K
TRAP
TS P
TRC
T0
LIQUID OUT
ADDING OR MULTIPLYING A CONSTANT
Concept #4
Know How to Design an Ethernet
Network for Speed and Reliability
Typical Network Architecture
Special Requirements for Industrial Control
• Industrial networks have a special requirements:
– Predictable throughput and scheduling
– Extremely low (zero?) downtime
– Scalable size (without performance loss)
– Used and maintained by people who are not communications specialists
– Operate in harsh and hazardous environments
– May impact safety if malfunctioning/not-functioning
• Industrial automation vendors originally developed their own (proprietary)
communication technologies
• Most automation systems today borrow technology from business/IT
networks and (possibly) modify it to meet industrial requirements
Instrumentation LANs
• Plants today use a lot of “smart” instrumentation, field devices and control elements
• This saves a lot of wiring and pushes control intelligence closer to the process
Instrument bus
Distributed I/O
Ethernet Signaling
• 10 Mbps (bus-topology) Ethernet uses a 1.2 V p-p Manchester Encoded signal.
Signal transitions rather than levels indicate data:
One = Negative Transition Zero = Positive Transition
• Advantage: Better RF immunity and built-in clocking
• Ethernet circuitry looks for two voltage levels:
– > 1.2 Volt pp: Signal on cable - network busy (don’t talk)
– > 1.5 Volt pp: Two Signals on cable - collision occurring (back off)
= 0
Signal on Wire
There is an ‘edge’
in the center of
each bit being
=1 transmitted
1 0 1 1 0 0 1 0
• 100 Mbps Fast (and Gigabit) Ethernet uses a different signaling scheme called MLT3 (much more
complicated)
Encapsulating the Data
• As a message passes down through the layers, each adds some information (a header) that will
be used by the corresponding layer at the receiving end.
• Industrial protocols tend to have few layers (three) in order to minimize the added overhead that
layering introduces
Program Sending Data Program Receiving Data
Data All the stuff below is transparent to the two applications Data
App Data App Data
Header Header
Pres Data Pres Data
Header Header
Sess Data Sess Data
Header Header
Trans Data TransH Data
Header eader
Netwk Data Netwk Data
Header Header
DL DL DL DL
Data Tail Header Data Tail
Header
Transfer Over Network
The “real” data
Physical The “overhead”
Port Numbers: TCP & UDP
• IP addresses get messages delivered to the right ‘host’ but some program in the source host is sending
those messages, and they are intended for a program running on the destination host.
• Port numbers are a 16 bit integer value (0-65,535) assigned to every program that requests/offers TCP or
UDP message delivery.
• Programs (‘services’) that are frequently used/needed, such as a web server, ftp, telnet, email, etc. have
permanent, fixed port numbers (0- 1024 are reserved for these ‘well known’ services)
• Clients requesting those services generally get assigned temporary (reusable) port numbers (> 1024) since
such programs come and go
Actual message delivery is from “port” to “port” (application to application)
37 5 21 81 6 94 62 4 1 66 3 91 41 8
SNMP
SMTP
SNMP
telnet
SMTP
HTTP
SNMP
telnet
TFTP
SMTP
HTTP
telnet
TFTP
HTTP
TFTP
NFS
FTP
NFS
FTP
NFS
FTP
TCP UDP TCP UDP TCP UDP
IP ICMP RIP ARP IP ICMP RIP ARP IP ICMP RIP ARP
PPP PPP ETHERNET ETHERNET
EIA-232 EIA-232 CAT- 5 CAT - 5
45.154.12.9 : 37 202.61.101.55 : 4
IP Supports Multiple Sessions
Application
Application
Application
Application
Port 1 Port N Port 1 Port N TCP/UDP
“Demultiplexed” based
on application (port#)
UDP protocol ICMP Protocol TCP protocol
IP
“Demultiplexed” based
on type of datagram
ARP Module IP Module RARP Module
Data Link
(Ethernet)
“Demultiplexed” based
Now on type of frame
Link layer protocol like
SNAP or RST
The past
Frames arrive sequentially in time on
the LAN (time-division multiplexed) Frame received Ethernet NIC
Concept #5
Know How to Effectively Manage
An Automation Project
Automation and Control Projects
• Big Opportunity
• Interface to Multiple Disciplines
• Lots of Hardware and Software
• Rapidly Changing Technology
• Real-Time Data Acquisition and Control, Possibly High-Speed
• Custom Software Development
• Intangible Deliverables
• High Customer Involvement and Expectations
• Integration Across Department Lines
And when Part of a Larger Project
• Usually involves an integration process with the enterprise
– Manufacturing production systems
– Information systems backbone
– Business process
• Involves standard models/Misc. sources of information
– ISA95
– ISA88
– Project Management Institute
– Construction Industry Institute
– Control System Integrations Association
• Automation and control work is done last so it tends to get a disproportionate
share of attention
Best Practice Project Phases
WHAT HOW DO
Design
Freeze
Project “Approval”
Mfg.
Feasibility Prep.
Definition Integrated Operate and
Testing, and
Customer
require-
Basis of
design
Detail Design
and
Demonstration Maintain
ments Project
Order Of
PM Planning Programming Operator
Commissioning Closeout
Training
Magnitude and Startup
75%
Definitive Long Term Support
Budget 10%
25%
Construction and
Checkout
Estimates
Major Facilitated
Reviews
time Mechanical
completion
Project Lifecycle — The ISA CAP Model
ISA Certified Automation Professional (CAP) Model
Questions?
Thank you for Your Participation
In Today’s Webinar!
