Professional Documents
Culture Documents
Gas Cracker
Reliance Industries Ltd
DMD
+
Column P
Pressure
Operator's
Preferred
Operating
Temperature Region
Purity
Valve Positions
APC Technologies
Commercially available technologies in APC:
− DMCplus (AspenTech)
− RMPCT (Honeywell)
− SMOC (Shell)
− STAR (DOT Products)
− CONNOISSEUR (Invensys)
APC Initiative at RIL
• STAR Controller
- Corporate License (1994)
- MEG-I (1995)
- PE (1996)
- PTA, Hazira & PG (1997)
- Polyester CP-2,3,4,7,8(1997-98)
• SMCA Controller
- Licensing (1994)
- PX (1995)
- LAB (1995)
APC Initiative at RIL
• DMCplus Controller
- Hazira Cracker (1998)
- Corporate License
(2001)
- Jamnagar Refinery
(2002…)
• APC Summary at RIL
MVs CVs DVs Controllers
• RIL, Jamnagar
− FCC Plant (1 Controller Remodeling for Revamp, Mar/06-Jun/06)
• RIL, Patalganga
- PX Plant (5 Controllers Nov/04 - Sept/05)
- PTA Plant (1 Controller Dec/05 - Mar/06)
- LAB Plant (2 Controllers Oct/04 – Sept/05)
- CP7 Plant (Started Feb/07)
• RIL, Nagothane
- Gas Cracker Plant (8 Controllers Jul/06)
- EO-EG Plant (Jul/06 - Jan/06)
• RIL, DMD
- Gas Cracker Plant (8 Controllers May/09)
APC Overview
Control Hierarchy
days Scheduling PIMPS+...
DMO / NOVA /
hrs Optimization
ROMEO
LP - CLP
min DMCplus /
STAR /
APC APC
CONNOISSEUR
DCS
msec
PLANT
Terms and Definitions
• Independent variables
• Dependent variables
• Unit step response curve
• Time to steady-state/settling time
Independent Variables (1)
• Independent Variable: A variable whose value is not affected
in any way by any other variable in the process
“Independence” C LC
to the A
FC
temperature controller A
Feedforward/Disturbance
Independent Variables (3)
TI PC CW
A
Temperature is
now dependent LC
A
FC
A
Independence promoted
to Flow Controller
Dependent Variables
• Dependent Variable: A variable whose dynamic behavior can
be described totally in terms of specific independent variable
changes over time
Trends
CV 1
MV1
STEP
MV 2
0.2
0.1
avg.
avg.
80 %
avg.
80 %
DMCplus Characteristics
• Multivariable
• Model-Predictive
• Constrained
• Optimizing
• Rigorous
DMCplus Characteristics:
Multivariable
• Changes in one independent variable (setpoint) effect
several controlled or constrained variables
• DMCplus controller response
– DMCplus accounts for the effects of all changes on all
controlled variables and finds the best overall solution
– DMCplus will find a control plan that will satisfy all
process and operating constraints, if such a plan
exists
Distillation Tower Schematic
OHTEMP OHPRESS
T P
MVs P CONDDP
FFs OHPVALV
OHLEVEL
REFLUX L
CVs OHPROD
F F
COLDP
P
DRAWT SSFLOW
SSCOMP
FEED FEEDT A 95%
T F
F T
REBBTU BTMFLOW
BTU F
Dynamic Model of Distillation
Tower
DMCplus Characteristics: Model-
Predictive
• Dependent variables will not remain at their current
values if independent variables (setpoints and
disturbances) have been moving
• DMCplus controller response
– DMCplus uses linear dynamic models to predict the
open-loop behavior of the controlled variables
• Predicts the violation of constraints and takes
control action based on the prediction
• Constantly updated with measured information
• Can model processes with unusual dynamics
DMCplus Characteristics:
Constrained
• There are constraints on independent and dependent
process variables
• DMCplus controller response
– All constraints are considered and accounted for in
the overall control plan
– High, low and rate-of-change limits on MVs
• Regulatory control loop setpoints or valve outputs
• Always satisfied
– High and low limits (setpoints) for CVs
• Product composition specifications, metallurgical
limits, valve output clamps
• Satisfied, if physically possible
DMCplus Overview: Optimizing
• Process should be operated in an optimal manner
• DMCplus controller solution
– Steady-State
• Maximize profit
• Robust handling of infeasibility
– Dynamic
• Minimize CV error
• Minimize MV movement
• Follow the optimum dynamic path to economically
optimum steady-state targets
APC Architecture
DMCplus Architecture
• Step 1- Read Process Values
• Step 2- Predict Steady State (SS)
• Step 3- Optimize at Predicted SS
• Step 4- Generate a Move Plan to reach SS
• Step 5- Implement Set points
Step-1: Read Process Values
Controlled Variable
Operating Limits
Steady
Manipulated Variable State
Current Values Linear Program
Manipulated Variable
Operating Limits
Controlled Variable
Steady State Targets
Economic Information
Step-3 Optimize at SS
• MV Parameters
- Cost
• CV Parameters
- Ranks
- Equal Concern Errors (ECE)
Step-4: Generate Move Plan
Step-4 Move Plan
• MV Parameters
- Move Suppression
- Maximum Move
• CV Parameters
- Equal Concern Errors (ECE)
Step-5: Implement Setpoints
• Check Data Quality
• Check Validity Limits
• Check Operator Service Status
• Write Setpoints
DMCplus Architecture
Economics
Costs on Manipulated Variables
Dynamic Predictions MV
Targets
Steady-State Dynamic
Prediction
Optimization Control
Module Steady-
State
Module CV Module
Predictions Targets
Dependent Setpoint
Variable Independent Manipulated Output
Values Variable Values Variable Values Values
ACOBASE / Cim-IO
Offline
Prod.
DMCplus
Online ACOBASE (Connect Cim-IO Client) Control
Context
Web
Web Cim-IO
Cim-IO Core (Cim-IO Kernel) Viewer
Server Server
- Upper validity limit attribute for all MVs will be read from the
DCS system and controller will not allow setpoint changes
beyond these limits
Control Philosophy (6)
• Software Failure
Write failure due to CIMIO or any other reason can be
used to trigger:
− Kill DMC process immediately or after one cycle
− Stop Controller
− Continue and let DCS watch dog stop the process
Control Philosophy (7)
• Hardware Failure
Prices
SS gain
Rank
Ethane_FM
Zone C2 C2 Level
CGC Cold
H10 Hot
3 Stg pr
rd
Constraint Constraint
Zone C3
H11 Propane_FM
C3 Level Total C2 + C3
H12
% Vol / Vol of C2
H13
H14
Variables (Ethane_FM)
MV (2) DV (1+3* 5 = 16) CV (3 + 3*5 = 18 )
H10/11/12/13/14 Zone 3 C3
flow
DMC Matrix for Ethane_FM(1)
DMC Matrix for Ethane_FM(2)
DMC Matrix for Propane_FM(1)
DMC Matrix for Propane_FM(2)
Variables (CGC)
MV (2) DV (6) CV (15)
Dummy_CGC Tot C2 FF PI3704PV
CGC 3rd Stage Disch Pres
• The product utilizes the same Linear Program (LP) technology that is
embedded inside the DMCplus controllers. The Composite LP variable set
is a superset comprised of every controlled variable (CV), manipulated
variable (MV) and disturbance variable (DV) that is contained in any
participating controller. Therefore a solution from the Composite LP,
honors the constraints and utilizes the manipulated variables of all the
participating controllers.