APC Training at Bina

Refinery, 26-30 August

2019

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
1
Day 1 Agenda
1. Morning Session:
 PID Control
 Introduction to APC

2. Afternoon Session:
 Process Modeling in PACE

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
2
PID Control

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
3
Block Diagram of a PID Control Loop

 The function of a PID controller can be understood through a block

or loop diagram:

SP E=CV-SP PID OP
Process
Control
Equipment
Algorithm

CVI
CV

CV (or, PV) = Control Variable (or, Process Variable)

SP (or, MV) = Setpoint (or, Manipulated Variable)

OP = Valve Opening

<PID is a Single Input Single Output (SISO) Controller>

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
4
PID Equations

{ 1
𝑀𝑉 ( 𝑡 )= 𝐾 𝑐 ∗ 𝐸 ( 𝑡 )+ ∫ 𝐸 ( 𝑡 ) 𝑑𝑡 +𝑇𝐷∗
𝑇𝐼
𝑑𝐸(𝑡)
𝑑𝑡 } Basic PID Eqn.

100 1
{ ()
𝑀𝑉 ( 𝑡 )= ∗ 𝐸 𝑡 + ∫ 𝐸 (𝑡 ) 𝑑𝑡+𝑇𝐷∗
NOTES:
𝑃𝐵 𝑇𝐼
𝑑𝐸(𝑡)
𝑑𝑡 } PID Eqn. in YIL DCS

• E(t) = PV(t) - SP(t), E = Error, PV = Process Value, SP = Setpoint

• As Kc (Proportional Constant) increases, or PB (Proportional Band, Range = <0:1000>)
Decreases, Proportional action increases
• As TI (Integral Constant, Range = <0.1:10000> sec) decreases, Integral action increases
• As TD (Derivative Constant, Range = <0:10000> sec) increases, Derivative action increases
• PID controller Equation is implemented in Discrete form
• There are other variants of Basic PID Equation, e.g., PV-Derivative or PI-D, and PV- Proportional
& Derivative or I-PD Equation.
• Tuning Methods used are Open-Loop (e.g., Ziegler-Nichols & Cohen-Coon), Closed-Loop (e.g.,
Ziegler-Nichols), Model-Based (e.g., IMC & ISE-IAE-IATE), and Trial & Error

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
5
Effect of Changing PID Parameters

PB=200

Kc=400

Kc=200, TI=2400, TD=120

Kc=200, TI=2400

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
6
Tuning Guidelines

Sl Proportional Integral Derivative

No Loop Type Scenario What is required?
Action Action Action

Accurate but stabilized

Setpoint tracking is
Response is fast.
required, as operator
1 Flow PV is generally Less More Nil
change Flow Setpoint
noisy.
regularly. Generally, PI
controller is used.
Stabilized control within a
Response is often desired range around
2 Level slow due to Setpoint is required. More Less Nil
inventory effect. Generally, P or PI controller
is used.
Response can have
some delay Tight but Stabilized control
3 Pressure depending on loop is required. Generally, PI More Medium Nil
design. PV is controller is used
generally noisy.
Response is Tight but Stabilized control
4 Temperature delayed due to heat is required. Generally, PID More Medium Less
transportation lag. controller is used

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
7
Block Diagram of Regulatory Control System in a
Process Equipment
 All PID or other Regulatory Controllers (e.e., Cascade, Ratio, etc.)
and PV Measurements in a process equipment can be represented
by the following block diagram:

MV1 CV1

MV2 CV2

MV3 CV3

DV1 CV4
Process
DV2 CV5

DV3
CV6

CV1, CV2 & CV3 = CV’s controlled by PID using MV1, MV2 & MV3 respectively

CV4, CV5 & CV6 = CV’s that are Uncontrolled

DV1, DV2 & DV3 = Disturbance variables that are Unaccounted

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
8
Drawback of Regulatory Control System – Summary

Multiple Input Multiple Output (MIMO) Interactions not handled

No Economic Optimization is possible
Improperly tuned PID parameters – operators often preferring to
keep the loop under manual mode!
Frequent Disturbances (such as variation in raw materials or
feed) and Frequent variation in Production objective makes
control more difficult
Safety & Hardware Constraints – Operators take very
conservative moves
Unavailability of on-line measurements – Product Qualities are
hard to control

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
9
 Drawbacks of Regulatory Control System

 Multi-Input Multi-Output 1 – Conventional PID controllers

3
(MIMO) Interactions 2.8
2.6
2.4
2.2 C3’ in PROPYLENE
2
1.8
1.6
Reflux Drum 1.4
1.2
PROPYLENE 1
0.8
0.6
F Q 0.4
C3’ 0.2
C3’’ in PROPANE
0
2 - Detune both controllers

C3’ in PROPYLENE
PP SPLITTER
C3’’ in PROPANE

F
3 – Detune only the less important controller

C3’ in PROPYLENE

Q C3’’
C3’’ in PROPANE
PROPANE

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
10
Drawback of Regulatory Control System

 Economic Optimization is absent

Lost Benefit

Optimum
Column dP
Reactor
Temperature
Benefit of moving

Compressor
speed
Pressure

Operator’s
Comfort-zone
it

Process Variable

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
11
Drawback of Regulatory Control System

 How Economic Optimization is possible

Old Standard Deviation

C Reduced
O variability alone
N does not achieve
New Standard Deviation big benefits....
S
T

Constraint
R
A
Operation
I
t closer to
r ofi N
P constraints
T
brings in the
big benefits …

Old S.P. New S.P.

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
12
Introduction to Advanced
Process Control (APC)

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
13
APC System Structure

 Generates MV’s for critical PID controllers, control a subset of both controlled
and uncontrolled CV’s, and uses all important DVs

MV1 CV1

MV2 CV2

MV3 CV3

DV1 CV4

Process
DV2 CV5

DV3
CV6

APC

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
14
APC Model Structure - Overall

 A supervisory controller that overcomes many drawbacks of regulatory control,

account for MIMO interactions and enable economic optimization

MV1 CV1
MV2

DV1 CV2

DV2

MIMO model used in APC

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
15
 APC Calculation Steps

 Every execution cycle

− Values of all APC controller input variables (including MVs, CVs, DVs,
Switch values, Calculation & Inferential inputs, etc.) are read from DCS.

− Dynamic model predictions along the output horizon are calculated based
on current values of MVs, DVs, Calculation & Inferential outputs, and the
difference of such predictions from the set ranges are calculated.

− Required actions (Delta MVs) along the input horizon are calculated by
minimizing an overall objective based of prediction errors, MV moves &
Economic objectives, subject to various output/input constraints

− Only the first step of MVs are implemented by writing to appropriate DCS
tags.

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
16
APC Advantages Over PID

The heart of the controller is a model for the entire process, hence the
alternate name MPC (Model Predictive Control) – the model captures all
a priori process knowledge
Rigorously account for MIMO interactions and safety/hardware
constraints
Account for process disturbances and variation in production objectives
Standard Deviation of control variables improves
Perform dynamic economic optimization and drive the process smoothly
towards the optimal operating point
Has tool to infer critical quality variables that cannot be measured on-line
– improves control of quality

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
17
APC Performance under MIMO Interactions

Reflux Drum

PROPYLENE
F Q
C3’ C3’ in PROPYLENE

PP SPLITTER

C3’’ in PROPANE
F

Q C3’’

PROPANE

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
18
Tangible & Intagible APC Benefits

Tangible Benefits of APC Project:

 Improvement in control of CV’s and reduction of their standard deviation
 Economic optimization:
 Increase of throughput or capacity utilization
 Increase in recovery or yield of desired product
 Reduction of energy/utility/operating costs

Intangible Benefits of APC Project:

 All PID are tuned to make the basic/DCS control more smooth and stable
 Less off-spec products or instability during throughput or feed or production objective changes
 Plant constraints are automatically managed provided they can be modeled
 Process become more stable, i.e., less process alarms are generated
 Operation staff need to set only the overall operational objective and constraints instead of setting the
parameters of several PID loops

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
19
Typical APC Project Activities

 Benefit Estimation Study (before PO)

 Kick-Off Meeting, followed by Detailed Process Study and
Pretest including Base Layer Control Review & PID Tuning
(after PO)
− APC performance is dependent on PID performance
− All APC related PID loops should be checked and tuned if
necessary
 Development of FDS (Functional Design Specification) or
preliminary controller design
 Design of Step Tests (i.e., specify the magnitude of steps to be
given to APC MVs)
 Execute Step Tests
− If signal/noise ratio is low (i.e., response is not distinct), step
test is to be repeated

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
20
Typical APC Project Activities (Continued)

 Identification of Controller Models using AIDA (an offline design

tool)
− Estimation of Model Parameters
− Model Validation
− Adequacy Check
 Development of Quality Inference Models using RQE (an offline
design tool)
− Estimation of Model Parameters
− Model Validation
− Adequacy Check
 Controller Development based on AIDA/RQE models and offline
simulation using SMOC (an offline design tool)
 Engineer Training & FAT (Factory Acceptance Test) based on
off-line simulation of SMOC for various test conditions/scenarios

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
21
Typical APC Project Activities (Continued)

 APC Graphics & Interface Development

− CV/MV/DV tables with entries necessary to run APC from DCS
are created
− DCS switches and logic are created that will enable safe run of
APC from DCS (for example, facility to switch off APC
following communication problem or plant emergency and
bumpless transfer from APC to full DCS control)
 Controller Commissioning & Operator Training
− Operators are given short training to get used to APC
operation from DCS
− Tuning of APC parameters (tuning parameters include model
and control algorithm related parameters including CV/MV
limits)
− Re-modeling is done if found necessary
− APC is commissioned with tight MV limits, which are gradually
relaxed to get optimization benefits

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
22
Typical APC Project Activities (Continued)
 Development of DDS (Detailed Design Specification) or final
controller design, and DCS Interface Design Document
 SAT (Site Activity Test)
− This is to prove that APC can run uninterrupted for ~15 days
when plant conditions are normal
 Post Implementation Audit
− This is to prove the APC economic benefits (based on data
from APC ON & OFF runs as well as base case data) and
percent confidence level of RQE predictions with respect to
Lab data
Warranty Period
− APC run beyond commissioning

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
23
Typical Post-Project Activities

Customer can buy AMC Contract that normally cover one or

more of following activities (alternatively customer can order each
service based on his need):
− Periodic site visit to evaluate APC working condition
− Tune APC for better performance (tuning parameters include
model and control algorithm related parameters including
CV/MV limits)
− Make minor design changes (e.g., CV/MV addition/deletion,
model parameter re-estimation) if necessary
− Re-audit APC benefit
− Attend to APC-related troubles using either remote phone
support or emergency site visit

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
24
Yokogawa APC Products

 Yokogawa APC products are outcome of the worldwide alliance

between Shell Global Solutions and Yokogawa since 1999.

 The alliance has unique strength by combining process know

how and plant automation experience resulting in sustainable
APC benefits delivered to customers.

 Before introduction of PACE, offline design was based on PCTP

(Process Control Technology Package), whereas Runtime control
& Inferential prediction were through ExaSMOC & ExaRQE
packages respectively.

 The new generation of Yokogawa APC product is called PACE

(Platform for Advanced Control & Estimation) package which has
separate versions for offline Design & Runtime use. It has been
developed with the objectives of getting better APC performance
and sustainability with easier implementation & maintainance.

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
25
The road to success means developing
sustainable solutions

| Document Number | March 23, 2016 |

© Yokogawa Electric Corporation
26