Refinery wide model:
London, June 2018
01Galp
Fingerprint
02
Galp
Refining
System
03Planning
04Scheduling
05Unit
Monitoring
06
07Questions
Refinery wide model: a real work example
Other
Applications
3
Galp fingerprint
330 kbpd
refining capacity
1 459
service Stations
0.6 million
clients
173 MW
cogeneration capacity
100 thousand
barrels of daily production
7
countries
PORTUGAL
SPAIN
EAST TIMOR
CAPE
VERDE
GUINEA-
BISSAU
S. TOMÉ
AND PRÍNCIPE
BRAZIL
ANGOLA
MOZAMBIQUE
NAMÍBIA
SWAZILAND
01Galp
Fingerprint
02
Galp
Refining
System
03Planning
04Scheduling
05Unit
Monitoring
06
07Questions
Refinery wide model: a real work example
Other
Applications
5
Galp Refining System
Matosinhos
Sines
• Total refining capacity: 330 kbd
• 100 % of Portuguese capacity;
• 20% Iberian Peninsula capacity Capacity: 110 kbd
Hydroskimmimg with Visbreaker unit
Lubs Plant
Aromatics Plant
Capacity: 220 kbd
High conversion with FCC and
Hydrocracker units
Our refining system
23%
Gasoline
38%
Gasoil
9%
Jet
16%
Fuel
2%
Others
7%
C&Q
2%
Gases
3%
Aromatics
2017
Galp Production profile:
6
Galp Refining System
Crude acquisition strategy
Latin America Middle East North Africa
FSU North Sea West Africa
26%
9%
17%7%
48%
2014/2015
49%
6%
18%
6%
18%
3%
2016/2017
7
Petro-SIM Refinery wide model
Plays a fundamental role in refining process optimization
Atmospheric distillation
Naphtha
hydrotreater
CCR (REF–SIM)
Vacuum distillation
Visbreaker (VIS–SIM) VGO hydrotreater (VGOHTR-SIM)
Hydrocraker (HCR-SIM)Diesel hydrotreater
FCC (FCC–SIM)
8
Petro-SIM Refinery wide model
Utilization scope
PLANNING
TROUBLESHOOTING
Accurate data to LP which guarantees the market needs and lead to the best economical
result
UNIT MONITORING
SCHEDULLING
Monitoring model prediction adherence to process unit real performance
Optimize cut-points for crude-mix feed and predict properties/yields for all units in order to
provide the integrated operational instructions to Operations.
Evaluate the impact of new or revamped units or in Galp margin
INVESTMENT
ANALYSIS
Analyzing equipment upsets and operating conditions alterations
01Galp
Fingerprint
02
Galp
Refining
System
03Planning
04Scheduling
05Unit
Monitoring
06
07Questions
Refinery wide model: a real work example
Other
Applications
10
Planning cycle
How information passes through
Raw materials
Operating
conditions
Product yields
Product qualities
REFINERYTRADING MARKET
Operating
restrictions
PLANNING
SIMULATION
MODEL
11
Petro-SIM Refinery wide model
Petro-SIM data to LP
1 . Node structure 2 . Vector structure
FEEDSTOCK
OPERATINGCONDITIONS
PRODUCTS
(YIELDS&QUALITIES)
 Discrete values;
 Heavy structure.
FEEDSTOCK
OPERATING CONDITIONS
PRODUCTS
(YIELDS & QUALITIES)
 Base case definition;
 Higher flexibility;
 Linearity problems.
01Galp
Fingerprint
02
Galp
Refining
System
03Planning
04Scheduling
05Unit
Monitoring
06
07Questions
Refinery wide model: a real work example
Other
Applications
13
CRUDE-MIX
OPERATING
CONDITIONS
OPTIMIZED CUT-POINTS
FEEDSTOCKS
YIELD AND PROPERTY
PREDICTIONS
Petro-SIM Refinery wide model
Schedulling
14
CRUDE-MIX
OPERATING
CONDITIONS
YIELD AND
PROPERTY
PREDICTIONS
INTEGRATED
OPERATIONAL
INSTRUCTIONS
Petro-SIM Refinery wide model
Schedulling
01Galp
Fingerprint
02
Galp
Refining
System
03Planning
04Scheduling
05Unit
Monitoring
06
07Questions
Refinery wide model: a real work example
Other
Applications
16
Unit monitoring
REFINERY
Petro-SIM UNIT MODEL
REAL YIELDS &
PRODUCT PROPERTIES
PREDICTED YIELDS AND
PROPERTIES
OPERATING
CONDITIONS
REAL
vs.
PREDICTION
• SISTEMATIC DEVIATION
REAL VS PREDICTION
• RECALIBRATE UNIT
MODEL;
• GIVE NEW INPUTS TO LP
SUBMODEL
• GOOD PREDICTION
• CALIBRATION OK
• GOOD INPUT TO LP
CRUDES/FEEDSTOCKS
PROCESS UNIT
FEED
(lab analysis)
Model validation and early intervention in calibrations
Calibration Criteria Standard: Systematic
deviation between reality and predictions
 2% wt for product yields;
 2% vol for recovered 360◦C in diesel.
17
Monitoring models prediction adherence to real units performance
Monthly bulletin
50
55
60
65
70
75
80
85
90
95
0
5
10
15
20
25
30
35
Jan-18 Feb-18 Mar-18
Reformateyield(wt%)
YieldsH2+FG,LPG(wt%)
CCR reforming unit
Objective:
 Detect early deviations between reality and simulation model
predictions. Early intervention in data inputs to LP.
 Identify if deviations between reality and LP are related to model
calibrations or to different feed and/or operating conditions;
Reformate H2+FG LPG
Real LP Simulation
with real feed
Real LP Real LP Real LP
0
10
20
30
40
50
60
PNAContent(%wt)
C6 C7 C8
C9 C10
Real
LP
Jan
101
101
Reformate
RON 3300
Feb
100
101
Mar
101
101
Paraffins
Naphtenes
Aromatics
18
Simulation model status assessement
Calibration Unit
Refinery:Sines
2017 Distillation unit
2016 Vacuum I
2016 Vacuum II
2016 Visbreaker
2015 Platforming
2017 Hydrocracker
2015 FCC
2017 Dessulfurisations
Calibration Unit
Refinery:Matosinhos
2017 Distillation unit
2017 Vacuum unit
2015 Visbreaker
2017 Reforming 1300
2015 Reforming 3300
2017 Aromatics plant
Monthly bulletin:
 Assess calibration of unit simulation models (LP submodels provided for all users)
and points out new calibration requirements.
19
Model Simulation management:
Integrated procedure for new calibration implementation in LP
Publish Simulation
Models Bulletin and
distributes to Refinery
and Planning Teams
Presents new calibration
proposals to Refinery
team and Planning team
Comment/approves
calibration proposals
Implements new
calibrations in the refinery
wide model or process
unit model and generates
data to LP.
Test new calibration and
implements in the LP.
SIMULATION
MODEL TEAM:
PLANNING TEAM:
SIMULATION
MODEL TEAM:
REFINERY
PERFORMANCE
TEAM:
01Galp
Fingerprint
02
Galp
Refining
System
03Planning
04Scheduling
05Unit
Monitoring
06
07Questions
Refinery wide model: a real work example
Other
Applications
21
Investment analysis
01
02
03
04
05
Licensor Data
Unit Calibration
Unit Validation
Refinery Wide Model
LP Model
FCC
MHC
VF
IX
FCC Cat Cooler
Mild Hydrocracker
Isomerization
Vacuum Flasher
LP studies with new or revamped units
22
Investment analysis
2019
1 price scenario
2020
4 price scenarios
2023
4 price scenarios 00
05
10
15
20
25
100 150 200 250 300
LPmargin(m$/y)
Unit critical spread
LP studies with new or revamped units
 Model implementation in the Refinery Wide Model;
 Utilization of different narratives for the refining
environment allows the analysis of the unit’s long-term
resilience.
0
5
10
15
20
25
30
35
40
45
Temperature(C)
Train A
Train B
23
Troubleshooting
Management of Crude Preheat Trains subject to fouling
Fouling; Factors;
Duties; €€€
Payback time – 26 days Cleaning optimal time – 360 days
0
2000
4000
6000
8000
10000
12000
14000
0 200 400 600 800 1000 1200 1400
Cost(k€)
Run time
Fuel Gas consumption
Cleaning Cost
Total Cost
tóptimo
RTDB
Historian
Connection
Petro-SIM
Explorer
01Galp
Fingerprint
02
Galp
Refining
System
03Planning
04Scheduling
05Unit
Monitoring
06
07Questions
Refinery wide model: a real work example
Other
Applications
Thank you for
the attention
Ana Rita Costa (rita.costa@galp.com)
Bernardo Barros (bernardo.barros@galp.com)
Cristina Ângelo (cristina.angelo@galp.com)
Cristovão Brandão (cristovao.brandao@galp.com)
Maria José Pinto (maria.jose.pinto@galp.com)

Europe User Conference: Galp refinery wide model - a real work example