Brenno C. Menezes Postdoctoral Fellow Technological Research Institute São Paulo, SP, Brazil Jeffrey D. Kelly CTO and Co-Founder IndustrIALgorithms Toronto, ON, Canada - Easy to implement (End-user as implementer: configure not code) - Integrates parts not yet integrated - Uses actual plant data - Reduces optimization search space in further problems (mainly MILP) - Tries to boost the polyhedral space of optimization (mainly NLP) - Automated-execution for faster and better solutions Smart: (six fundamentals) Smart Process Operations in Fuels Industries: Applications and Opportunities ITAM, Mexico City, Feb 5 th , 2016. TÉCNICAS AVANZADAS DE OPTIMIZACIÓN PARA EL SECTOR PETROLERO
37
Embed
Smart Process Operations in Fuels Industries: Applications and Opportunities (presentation)
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Brenno C. Menezes
Postdoctoral Fellow
Technological Research Institute
São Paulo, SP, Brazil
Jeffrey D. Kelly
CTO and Co-Founder
IndustrIALgorithms
Toronto, ON, Canada
- Easy to implement (End-user as implementer: configure not code)- Integrates parts not yet integrated- Uses actual plant data- Reduces optimization search space in further problems (mainly MILP)- Tries to boost the polyhedral space of optimization (mainly NLP)- Automated-execution for faster and better solutions
Smart: (six fundamentals)
Smart Process Operations in Fuels Industries:
Applications and Opportunities
ITAM, Mexico City, Feb 5th, 2016.
TÉCNICAS AVANZADAS DE OPTIMIZACIÓN PARA EL
SECTOR PETROLERO
Decision-Making Tools in the oil-refining industry
2
Space
Time
Supply Chain
Refinery
Process Unit
second hour day month year
RTOControl
on-line off-line
Scheduling
Operational Planning
Tactical Planning
Strategic Planning
Operational Corporate
week
Decision-Making Tools in PETROBRAS (in oil-refining)
3
Space
Time
Supply Chain
Refinery
Process Unit
second hour day month year
RTOControl
on-line off-line
Operational Planning
Tactical Planning
Strategic Planning
SimulationPetrobras
NLP Optimization Commercial (Aspentech)
LP Optimization Petrobras
Operational Corporate
week
Scheduling
4
Space
Time
Supply Chain
Refinery
Process Unit
second hour day month year
RTOControl
on-line off-line
Operational Planning
Tactical Planning
Strategic Planning
Simulation
NLP Optimization
LP Optimization
Operational Corporate
week
Decision-Making Tools in the oil-refining industry: Challenges
Scheduling
5
Space
Time
Supply Chain
Refinery
Process Unit
second hour day month year
RTOControl
on-line off-line
Operational Planning
Tactical Planning
Strategic Planning
Operational Corporate
week
1st: Logic variables (MILP)
2nd: Nonlinear Models (NLP)Optimization
(MILP and NLP)
ExpansionInstallation
Modes of operation
MaintenanceCleaning (Decoking)Catalyst Change
Integrations
time
space
Scheduling
Decision-Making Tools in the oil-refining industry: Challenges
(Menezes, Kelly, Grossmann & Vazacopoulos 2014)
(Menezes, Kelly & Grossmann, 2015)
(Kelly & Zyngier, 2016)
6
Space
Time
Supply Chain
Refinery
Process Unit
second hour day month year
RTOControl
on-line off-line
Operational Planning
Operational
week
Optimization(MILP and NLP)
(LP)
Integrations
time
space
Scheduling
Smart Process Operations: Applications around Scheduling
TFurnace Yields & Properties = base+delta x TFurnace
LN 20
Pinto et al, 2000; Neiro and Pinto, 2004
CDU/VDU Cut and Swing-Cut IBP (ºC) FBP (ºC)
Fuel Gas C1C2 -273 -50
LGP C3C4 -50 20
Light Naphtha LN 20 150
Heavy Naphtha HN 150 190
Kerosene K 190 250
Light Diesel LD 250 390
Heavy Diesel HD 390 420
Atmosferic Residue ATR 420 850
Light Vacuum Gasoil LVGO 420 580
Heavy Vacuum Gasoil HVGO 580 620
Vacuum Residue VR 620 850
150
CDU/VDU Cut and Swing-Cut TIB (ºC) TEB (ºC)
Fuel Gas C1C2 -273 -50
LGP C3C4 -50 20
Light Naphtha LN 20 140
Swing-Cut 1 SW1 140 160
Heavy Naphtha HN 160 180
Swing-Cut 2 SW2 180 210
Kerosene K 210 240
Swing-Cut 3 SW3 240 260
Light Diesel LD 260 360
Swing-Cut 4 SW4 360 380
Heavy Diesel HD 380 420
Atmosferic Residue ATR 420 850
Light Vacuum Gasoil LVGO 420 580
Heavy Vacuum Gasoil HVGO 580 620
Vacuum Residue VR 620 850
SW1
SW2
SW3
SW4CrudeA
Component/Psedocomponent(Micro-cuts)
BoilingPoint(ºC)
Yields(Vol%)
Gravity(Kg/m3)
Sulfur(W%)
Methane (CH4) -161.52 0.0041
Ethane (CH2-CH2) -88.59 0.0081
: : : :
N-pentane 36.09 0.0152
Hypo40 40 1.1427
Hypo50 50 1.4874
: : :
Hypo840 840 0.2544
Hypo850 850 0.0210
PT
CrudeB
CrudeC
CrudeD
K=y/x=Pvap/P
KHypo=f(P,T,Column)
Molar & EnergyBalances
Fixed Yields
Swing-Cuts
Delta Base
Chronology
LN
HN
SW1160
LN
140
Zhang et al, 2001; Li et al, 2005
20
Menezes, Kelly and Grossmann, 2013
CrudeA
Component/Psedocomponent(Micro-cuts)
BoilingPoint(ºC)
Yields(Vol%)
Gravity(Kg/m3)
Sulfur(W%)
Methane (CH4) -161.52 0.0041
Ethane (CH2-CH2) -88.59 0.0081
: : : :
N-pentane 36.09 0.0152
Hypo40 40 1.1427
Hypo50 50 1.4874
: : :
Hypo840 840 0.2544
Hypo850 850 0.0210
PT
CrudeB
CrudeC
CrudeD
K=y/x=Pvap/P
KHypo=f(P,T,Column)
Molar & EnergyBalances
Fixed Yields
Swing-Cuts
Fractionation Index (FI)
Delta Base
ChronologyAlattas, Grossmann and Palou-Rivera, 2011, 2012
Defines crude diet based on assay for Tcut=(IBPi+FBPj)/2
Defines new IBPi and FBPi
for a selected crude diet
CrudeA
Component/Psedocomponent(Micro-cuts)
BoilingPoint(ºC)
Yields(Vol%)
Gravity(Kg/m3)
Sulfur(W%)
Methane (CH4) -161.52 0.0041
Ethane (CH2-CH2) -88.59 0.0081
: : : :
N-pentane 36.09 0.0152
Hypo40 40 1.1427
Hypo50 50 1.4874
: : :
Hypo840 840 0.2544
Hypo850 850 0.0210
PT
CrudeB
CrudeC
CrudeD
K=y/x=Pvap/P
KHypo=f(P,T,Column)
Molar & EnergyBalances
Fixed Yields
Swing-Cuts
Fractionation Index (FI)
Delta Base
Chronology
Hybrid Models
Sanchez and Mahalec, 2012
Defines crude diet based on assay for Tcut=(IBPi+FBPj)/2
Component/Psedocomponent(Micro-cuts)
BoilingPoint(ºC)
Yields(Vol%)
Gravity(Kg/m3)
Sulfur(W%)
Methane (CH4) -161.52 0.0041
Ethane (CH2-CH2) -88.59 0.0081
: : : :
N-pentane 36.09 0.0152
Hypo40 40 1.1427
Hypo50 50 1.4874
: : :
Hypo840 840 0.2544
Hypo850 850 0.0210
PT
CrudeB
CrudeC
CrudeD
K=y/x=Pvap/P
KHypo=f(P,T,Column)
Molar & EnergyBalances
Fixed Yields
Swing-Cuts
Fractionation Index (FI)
Delta Base
ChronologyDBCTO
Hybrid Models
Kelly, Menezes and Grossmann, 2014
CrudeA
Defines new IBPi and FBPi
for a selected crude dietDefines crude diet based on assay for Tcut=(IBPi+FBPj)/2
Distillation Blending and Cutpoint Temperature Optimization (DBCTO)
From Other
Units
From CDU
Kerosene
Light Diesel
ATR
C1C2
C3C4
N
K
LD
HD
Naphtha
Heavy DieselCrude
CDU
ASTM D86
TBP
Inter-conversion
Evaporation
Curves
Interpolation
Ideal Blending
Evaporation
Curve
Multiple
Components
Final
Product
ASTM D86
Interpolation
Inter-conversion
TBP
(Kelly, Menezes & Grossmann, 2014)
Cutpoint Temperature Optimization
T01 T05 T10 T30 T50 T70 T90 T95 T99
Temperature
Yie
ld (
%) Back-end:
Front-end:
New Temperature: NT
Old Temperature: OTNew Yield: YNT
Cutpoint Temperature Optimization
Temperature (oF)
Yie
ld (
%) Back-end:
Front-end:
T99: 230→214
T01: 91→85
New Temperature: NT
Old Temperature: OTNew Yield: YNT
95.87%
-1.45%
Curves Renormalization
Difference in Yield: DYNT
Temperature (oF)
Yie
ld (
%)
Optimized (renormalized)
DYNT99
DYNT01
• Maximize flow of DC1 and DC2 ($0.9 for DC1 and $1.0 DC2) with lower and upper bounds of 0.0 and 100.0 m3 each (DC3 and DC4 are fixed at 1 m3). The ASTM D86 specifications are D10 ≤ 470, 540 ≤ D90 ≤ 630 and D99 ≤ 680.
Table. Inter-Converted TBP (ASTM D86) Temperatures in Degrees F.
Example
The new and optimized TBP
curve for DC1 given its front
and back-end shifts is now:
[(1.053%,312.8),
(10.015%,432.9),
(31.188%,521.6),
(52.361%,565.3),
(73.534%,606.4),
(94.707%,668.3),
(98.995%,689.3)]
Temperature (oF)
Yie
ld (
%)
Figure. ASTM D86 distillation curves, including the final
blend, which is determined by the blended TBP
interconversion to ASTM D86.21
Example
Solver: SLPQPE_CPLEX 12.6
Reduction in DC1’s T99 (TPB) from 715.7 to 689.3 oF
630 (ASTM D86)
Data-Driven Real-Time Optimization (DDRTO)
• Uses LP coefficients estimated directly from the plant or process using off-line closed-loop data and then we optimize this in real-time using an LP.
• Sits above a MPC layer to reset its targets or setpoints over time.• Optimizes IV’s subject to lower and upper bounds on both the IV’s and DV’s.
IVi is ith independent variableDVj is jth dependent variablewDVj is jth dependent variable’s profit or economic weight: cost (-) and price (+)bIVi and bDVj are the biases due to measurement feedbackSSGj,i is the steady-state gain elementPIVi and PDVj are the past valuesLIVi,UIVi, LDVj, UDVj are the lower and upper bounds.DLIV,i, DUIV,i and DLDV,j, DUDV,j are the lower and upper delta bounds.
• Steady-State Detection (SSD)- Determines if unit-operation is stationary (no accumulations) or at steady-state.
• Steady-State Data Reconciliation (SSDR)- Determines if unit-operation’s measurement system is statistically free of gross-errors and that there are no detectable losses/leaks.
• Steady-State Gain Estimation (SSGE)- Determines steady-state gains (actual first-order partial derivatives) using open- or closed-loop routine operating data though other hybrid methods such as rigorous models can be combined.
• Steady-State Gain Optimization (SSGO)- Determines new setpoints using quantity optimization and on-line measurement feedback to help “incrementally” situate the plant/sub-plant to a more profitable operating or processing space.
Data-Driven Real-Time Optimization (DDRTO)
Steady-State Gains (SSG’s)
Independent Variables (IV’s)
Dependent Variables (DV’s)
Data-Driven Real-Time Optimization (DDRTO)
IMPL’s UOPSS Visual Programming Language using DIA
Variable Names:
v2r_xmfm,t: unit-operation m flow variable
v3r_xjifj,i,t: unit-operation-port-state-unit-operation-port-state ji flow variable
v2r_ymsum,t: unit-operation m setup variable
v3r_yjisuj,i,t: unit-operation-port-state-unit-operation-port-state ji setup variable
VPLs (known as dataflow or diagrammatic programming) are based on the idea of "boxes and arrows", where boxes or other screen objects are treated as entities, connected by arrows, lines or arcs which represent relations (node-port constructs). (Bragg et al., 2004)
6- Differential Equation Solution (ODE and PDE): gProms
Applications in IMPL
Smart Operations: Opportunities in “Bottleneck” Scheduling
Step 1: Identify Key Bottlenecks (see below)
Step 2: Design Optimization Strategy
Step 3: Determine Information Requirements
Step 4: Prototype and Implement, etc.
Quantity-related:
Inventory containment Hydraulically constrained
Logic-related (Physics):
Mixing, certification delays, run-lengths, etc. Sequencing and timing
Quality-related (Chemistry):
Octane limits on gasoline Freeze and cloud-points on kerosene and diesels, etc
Step 5: Capture Benefits Immediately
(Harjunkoski, 2015)
Scheduling Solution Development Curves
Smart Process Operations: Opportunities in ICT
(Qin, 2014)(Christofides et al., 2007)
(Davis et al., 2012)
(Huang et al., 2012)
(Chongwatpol and Sharda, 2013)
(Ivanov et al., 2013)
Smart Process Manufacturing Big Data RFID in APS and Supply Chain
Example: when crude is selected for 2-4 days, after the 1st shift of 8h update all data usingInformation and Communication Technologies (ICT) integrated with Data-Mining applicationsand then use this in the Decision-Making.
36
Integration Strategies for Multi-Scale Optimization in the Oil-Refining
Industry (Multi-Layer and Multi-Entity)Brenno C. Menezes, Ignacio E. Grossmann and Jeffrey D. Kelly
reduce bottleneck and idling of equipment
maintenance of equipment
Operational Planning
Strategic Planning
Plants Terminals Fuels
Processing Distribution Marketing & Sales
Raw Material
Procurement
expansion, installation, extension of equipment
production level, supply chain service
Instrumentation, Advanced Process Control and RTO
Scheduling
Tactical Planning
model data
model data
model data
model data
cycle data orders (feedforward)
key indicators (feedback)
coordination and collaboration
modes of operation,campaign
on-line
off-line
Gracias
It seems a paradox, but I have been saying that the biggest human fear is not the fear of the darkness, but the fear of the light.