1. PredictPro Delivers Results at
Rompetrol’s Constanta Refinery
Crude Fractionation Unit
Radu Iscovici
Rompetrol
Chibuike Ukeje-Eloagu
Emerson Process Management

2. Agenda
Introductions (Presenter, Rompetrol, Emerson Team)
Project Execution
– Project Team
– Project Implementation
CDU Control Objectives & APC Scope
Control Design, Implementation & Commissioning
Benefits Audit
Long-Term Operability
Summary

3. Main Presenter
Radu Iscovici
Head, Process Control Department at Rompetrol Refining
SA, Romania
In charge of:
profit increase through advanced regulatory controls
and advanced process control and optimization
techniques
safe and secure plant operation through safety
instrumented systems implementation;
data collection, transfer and processing into valuable
information for management through implementation
of IT industrial systems

4. Rompetrol – An Overview
A leader in the regional oil industry
Rompetrol is a multinational company with activities
in 13 countries
Strong operational base in the Black Sea and
Mediterranean Sea region
The number and the complex nature of the projects
executed at a global level have turned Rompetrol
over time into the best known international brand
originating in Romania.
Co. Size: 9,000 Employees
Industry: Oil & Energy
Website: http://www.rompetrol.com

5. Constanta Refinery – A brief history
Built in the 1980’s for the Romanian National Oil
company
Now owned by Rompetrol
Largest Romanian refinery (100,000 bbl/day)
Recently modernized with Emerson’s DeltaV DCS
Continuous drive to improve operational efficiency
through exploitation of new DCS functionality and
smart field instrumentation

6. Emerson Process Management
Strategic Services Group
UK Based (Middlesbrough)
Support for Europe. Middle East &
Africa
Consultancy in advanced control
and optimization
Process experts with 10 to 30
years experience (ex operating
companies)
Experienced project managers
Process-specific technical solutions
(oil & gas, refining, chemicals,
utilities)

7. Project Execution
Project Team
Emerson
Emerson Romania
Strategic Services (UK)
AAT (US)
Rompetrol
Rompetrol Engineers embedded in the execution team
Renata-Alina Orzeiu (Control Engineer)
Gheorghe Oprea (Control Engineer)
Sebastian Paul Iftene (DCS Administrator)
This strategy ensures greater understanding of the
installed system thus developing on-site champions of the
APC technology
Any DCS configuration work carried out by Rompetrol
personnel

8. Project Execution
Project Implementation
Kick-off
Kick-off meeting to introduce team and review project
admin and QA
Site survey for data gathering & Preliminary Step testing
Design
Functional, Detailed & Acceptance Test Specifications
developed by Emerson for review/acceptance by
Rompetrol
Build
Build controllers, Step Test MV/DV, generate models
Commissioning
Commission controllers, deliver training, SAT.
Perform APC Benefits Audit

9. Project Execution
Major Milestones
June 2008 – Project kick-off
On-site kick-off, site survey & Preliminary Step testing
July 2008 – Commence Controller Build
Designs completed, commence site work
Controllers were built and commissioned unit by unit
Jan 2009 – Technical Completion
All controllers commissioned and in full use

10. CDU Control Objectives and Scope
1. Maximize Gas
Throughput
2. Maximize value of
product recovery
3
Naphtha
3. Maximize heat To Vacuum System
2
Recovery Steam
Kerosene
Steam
4. Maximize heater
efficiency
Lt Diesel
Steam 3
LVGO
Diesel
Steam
AGO
HVGO
4 Stripping Tower
1 4
Crude Feed
Steam
Steam
Residue
Fired Heaters Atmospheric Tower
Fired Vacuum Tower
Heater

11. CDU Control Design
Why Use MPC? P Gas
P
Interactions T
Disturbances F F F
Naphtha
Long Time T
FBP
T
To Vacuum
to Steady L
Steam
F
System
State
F
F F Kerosene
T
Steam
Optimization L F IBP FBP
F Lt Diesel
T Steam
L F IBP
O2 CO F
T F LVGO
T P F Diesel
F Steam
F T T F D360 Cong. Flood
L
F T
Flood F AGO
T F HVGO
F D360
Fuel HV Stripping
F O Flash
Tower
Fired Heaters O2
T F T P
O2 CO
T P Steam T
F T T
F
Crude F T
Steam
F T
Feed F T
F T
F T
Residue
Atmospheric
F F
Tower F
Vacuum
Air HV Fuel
HV Fuel Tower
Fired Heater

12. CDU Control Design
APC 4
(11 x 11) P Gas
P
T
F F F
Naphtha
APC 1 FBP
APC 5 (6x6)
T T
(3x3) O2 CO
Steam
T P L F
F
F T F F Kerosene
T
F T Steam
L F IBP FBP
F F Lt Diesel
Fuel HV
T Steam
L F IBP
F
T F LVGO
F Diesel
F Steam
T F D360 Cong. Flood
L
APC 2 (4x4) Flood F AGO
T F HVGO
D360
Stripping
O2 CO
F Tower APC 3 (5x5)
T T P O Flash
O2
F T F T P
Crude F T Steam T
F T T
F
Feed F T
Steam
F F F T
Fuel
Air HV F T
Atmospheric
Tower F
Vacuum
Fired Heaters HV Fuel Tower
Fired Heater

13. CDU Control Design (Atm Section)
Control Top C2 Base Base 4 Pressure 4 Product
Variables Pressure Level Steam Corrected Flow Rates
C1: MPC4 Ratio Temperatures
RCAS RCAS RCAS
AUTO AUTO AUTO
Control Manipulated
Variables Variables
dT O2 Outlet Outlet O2 dT Total
Temperature Temperature C1 3 Pump Around Flow
Flow Rates
H1: MPC1 H2: MPC2
4 Product Flow Rates
Reflux Drum Pressure
Pass Pass Fuel Pass Pass Fuel Air Flow
Manipulated 1 Flow 2 Flow Pressure 1 Flow 2 Flow PressureRate Base Steam Ratio
Variables Rate Rate Rate Rate

14. CDU Control Design (Vacuum Section)
Control C1 Base Average Feed 4 Pressure
Variables Level Temperature Corrected
C5: MPC5 Temperatures
RCAS RCAS
AUTO AUTO
Manipulated
Variables
Control Total C5 Outlet
Feed Temperature Recycle Flow
Variables
Flow
H3: MPC3 2 Pump Around
Flows
Quench Flow
Manipulated 4 Pass Fuel
Variables Flow Pressure
Rates

15. Inferential Analysis
Product compositions are not measured on-
line but once per day in the Laboratory.
Product compositions are key constraints in
a crude unit that dictate how much product
can be made.
A full suite of “virtual analyzers” was created
that predict the product properties (such as
diesel pour point) in real-time.
These virtual analyzers use a regression
based on product flow rates and column
temperatures.
DeltaV Neural is a standard component in
the DCS.
The predicted values are updated daily
against the laboratory results to ensure that
the predictions remain accurate.
These predicted values are used by the
controller as constraint variables to keep
products on specification.

16. Controller Configuration
Emerson’s PredictPro™
DeltaV™ standard component
Rompetrol DCS engineer configured 5 PredictPro™
Applications to Emerson design using Control Studio™

17. User Interface Configuration
PredictPro standard interface –
self configuring

18. Step Testing
Automatic Step Testing

19. Commissioning (1)
Model Generation:
– automatic
– manual adjustment
Model Verification
– by comparing Predicted vs.
Actual
Controller tuning
– fine tune response by
weighting CVs and MVs

20. Commissioning (2)
5 optimization
scenarios configured
Can assign values to
variables
Controller will
maximize total value
Use to maximize
production, and
minimize energy use
Operator selectable
in real-time

21. Benefits Audit (Summary)
Compare before and after operating
periods
Crude throughput constrained by
downstream units!
Increased production of more valuable
products for a given feed rate
Improved heater efficiency by a
significant amount
Quantified benefits:
– Audited project benefit of US $1.8m
(3-month payback)
Non-Quantified Benefits:
– Reduced operator actions
– Better response to disturbances such as
crude changes

22. Benefits Audit (Details)
Quantified benefits (Column):
– Reduced giveaway of valuable product
into AGO stream
– Quantified benefit:
• US$ 1.79 million / year
Quantified benefits (Heaters):
– Improved efficiency due to
• Pass balancing
• Improved excess O2 control
– Quantified benefit:
• US$ 103,000 / year
Payback Time:
– 3 months

23. Long-Term Operability
Since commissioning, the application has been in continuous
use.
12 month warranty for the APC application is included as part
of Emerson service.
Since entire application is constructed from standard DCS
components, the standard Emerson DeltaV DCS support
service applies.
– including 24/7 help line
– agreed man-hour rates if required
How regularly will model refits be required?
– Once models fitted they are tolerant of inaccuracies to a
reasonable degree.
– So long as no major process modifications are made the models
will not require refitting (e.g. changing the tuning of a control loop,
replacing a valve or a pump or cleaning a heat exchanger should
not require refitting)
– If models do require fitting, automatic tools make the process
quick and easy.

24. Summary
Total APC Audited
Benefit of US$ 1.8
million
Payback time of just
Plant Type: under 3 months
Refinery Familiar DCS
environment made
for a quick and
Scope: smooth uptake of the
Model Predictive Control on:
Atmospheric Column APC controllers by
Feed Heaters (x2) the operations crew
Vacuum Column
Vacuum feed Heater

25. Thank You
Comments / Questions?