Escolar Documentos
Profissional Documentos
Cultura Documentos
What Next?
Rafiqul Gani
CAPEC, Department of Chemical Engineering,
Technical University of Denmark, DK-2800 Lyngby, Denmark
&
Ignacio E. Grossmann
CAPD, Department of Chemical Engineering
Carnegie Mellon University, Pittsburgh, PA 15213, U.S.A.
Nutrition
Fertilizers Electronics Communications,
Entertainment
Vitamins Fuels
Mobility
Pharmaceuticals Plastics
DetergentsColorants and
Healthcare coatings
Clothing
Adapted from
Housing
Ploecker, 2005
Macro-Processing:
XA
XB
XC
Batch or Continuous Chemical Plants
XD
XE
XF
XG
XH
XD
XE
XF
XG
XH
Micro-Processing:
Plant-on-a-Chip
Molecular-Processing:
The Cell
Metabolites
9
Scope of the basic “products” of PSE/CAPE
Problem
Information
(knowledge, Design
Process/ data)
Product
Production
Operation Simulation
Planning
Environment
Model Analysis
Business
System
Plenary Lecture, ESCAPE-17, Bucharest, Romania, 27-30 May 2007 12
Computer aided modelling framework
Customized
Modelling tools: model reuse;
model aggregation/decomposition;
simulator
identification; analysis ….
CAPE-OPEN standards; plug & play
Modelling Models
Computer Numerical
science, IT analysis
Process-
Product
Thermal Interface
Raw Reactor/ Materials
Crystallizer CVD Reactor
materials Homogenizer
CNT SCF
Chromatography
Extruder Polymer Film
Blowing
C60
Food Packaging
Films
16
Diamondoid
Natural Herbs to Healthcare Products Ka Ng (2006)
Supercritical
Reaction/
SLE
Crystallization
VLE
Granules
Raw materials
Tablets
Supercritical
Granulation
Fluid Extraction Diffusion
Tester
Transdermal
patches
Supercritical
Fluid
Precipitator
17
I. Product and Process Synthesis, Design, Analysis
„Nowadays, chemical products are not sold for what they are
but, for what they do“
In-use properties
Process development
Product development
Product function
and manufacturing
Physico-chemical properties
Process function
Process
M. Kind, 2005
Pi = Ai + Bi + Ci + Di
Property
Models
(simple small molecules
& mixtures)
Electrolyte
Usually time
A lot of gaps in consuming and
the parameter VLE, SLE, LLE expensive and
matrix for the experiments to may not be
UNIFAC model! collect data possible
Gonzalez et al. A method for prediction of UNIFAC group interaction parameters, AIChE J (2007)
Process-product
synthesis/design problem algorithm/solution strategy
definition Process-product
behavior model
no
algorithm/solution strategy
Define design
Identify best match targets
of design targets?
yes Process-product
property model
Solution found
property models
high accuracy
low application range
MILP
Global Optimization
23
De Novo Protein Design (Chris Floudas, Princeton)
CAMD CAFD
Given a set of target properties,
find the molecules or mixture that Definition problem Given a multicomponent mixture,
match the target properties separate into a pure product streams
Generation
of alternatives
& optimal design
(iAD)(rAD/pABCDE)<1<2(fAB/ABCDE)1[(AB/CDE)(oAB)](C/DE)[oC)](D/E)2(oE)
Aqueous-phase
chemistry SECONDARY
POLLUTANTS
Reactions
Dispersion
Acid
Rain ATMOSPHERIC PRIMARY
PARTICLES POLLUTANTS
EMISSIONS
Sustainable
Integrated
Systems
Ethanol Fuel
Protein
Biochemical
(Gluten meal)
conversion
Succinic Acid
THF, BDO
Acrylic, PDO
3-HP Acid
Corn
Starch Sugars
Corn Mill
Food and
pharma Sorbitol Isosorbide
Glycerol
EG Replacement
Thermo-chemical
Fiber
conversion
Fatty Acid Olefins
Stover
9-deceonic acid
Thermo-chemical Energy
Lignin conversion Biodiesel
Stover Mill
Plenary Lecture, ESCAPE-17, Bucharest, Romania, 27-30 May 2007 29
Energy Optimization of Corn-based Bioethanol
Peschel, Martin, Karuppiah, Grossmann, Zullo, Martinson (2006)
Base case
Total equipment cost = 50.3 M$ Steam cost = 15 M$ /yr Manuf cost = 1.61 $/gal
Plenary Lecture, ESCAPE-17, Bucharest, Romania, 27-30 May 2007 30
Superstructure of a Bio-ethanol Plant
Required operations
Solids drying
Ethanol purification
Mechanical press before/after Beer column
Molecular Sieves vs Corn Grits
Cooling
Multieffect columns:
Low Pressure
column
High Pressure
column
Heat integration
Process streams
Total equipment cost = 49.7 M$ Steam cost = 4.8 M$/yr (-65%) Manuf cost = 1.43 $/gal
Reboiler Reboiler
Condenser
1.94 atm
Reboiler
1 atm
Condenser
Condenser
Rectification Column
Single column Double effect column
Reboiler Reboiler
1.97 atm
Condenser
Reboiler
Condenser
0.625 atm
Condenser
Pressure on process industry for efficient use of water and disposal of wastewater
GOAL:
GLOBAL OPTIMIZATION method for INTEGRATED DESIGN
of Water Using and Water Treating Units
(1 ppm A,
1.16 ppm B)
into environment
Freshwater
60 ton/hr PU3
M1
M1 PU1
PU1 S2
S2 M3
M3 TU1
TU1
TU1 S4
S4
S4
Freshwater
Freshwater
S1
S1 M5
M5
Discharge
Discharge
M2
M2
M2 PU2
PU2
PU2 S3
S3
S3 M4
M4
M4 TU2
TU2
TU2 S5
S5
S5
2. Splitters (SU) S
TU
4. Treatment Units (TU)
To M2
3
M
To
To
M
1 6
To M
M1 PU1 S2 M4 TU1 S5
1
M
M2
To
To
Freshwater Discharged
S1 M2 PU2 S3 M5 TU2 S6 M7
into environment
To
M
3
M3 PU3 S4 M6 TU3 S7
To
M4
2
1
To
M
M
To
M
To
3
Superstructure is formulated as a Non-Linear Programming (NLP) problem (Bilinear)
and solved with novel global optimization algorithm
Key features:
-Integrate strategic, tactical and operational decision-making
-Integration of the information, modeling and solution methods
Scope
Strategic Optimization
Modeling System Analytical Strategic Analysis
IT
Tactical Optimization Long-
Long-Term Tactical
Analysis
Modeling System
Demand Short-
Short-Term Tactical
Production Planning Optimization Logistics Optimization
Forecasting and Order Analysis
Modeling Systems Modeling System
Management System
Operational
Production Scheduling Distributions Scheduling Optimization
Analysis
Optimization Modeling Systems Modeling Systems
Transactional IT
Materials Requirement Distributions Requirements
Planning Systems Planning System
Enterprise Resource
Planning System
External Data
Management Systems
Targets
Pre-
Hits Submission& Lifecycle
clinical Phase 1 Phase 2a/b Phase 3
Leads Approval Management
Development
Candidate
2-5 yrs 0.5 - 2 yrs 1 - 2 yrs 1.5 - 3.5 yrs 2.5 - 4 yrs 0.5-2 yrs 10-20 yrs
Colin Gardner (Transform Pharmaceuticals)
Production Scheduling
Goal: Improve the asset utilization of geographically distributed assets and reduce
cost to serve by improving enterprise wide tactical production planning.
Production Plant
Customer
days, weeks
Scheduling
Scheduling
Challenges:
Challenges:
Objective
• Max Profit = Sales –Operating Costs – Inventory Costs – Transition Costs
44
Proposed Decomposition Algorithm
Yes
No STOP
No UB – LB < Tolerance ? Solution = LB
Storage Charging
Vessels
Crude-Distillation Unit
Uncertainty
Time
(1) (2)
¾ Small FPSO, converted from oil tanker Cannot produce oil, only drilling and
¾ Low capacity oil recovering capability
¾ Costs less
¾ Short construction time
(3) (4)
¾ The aim is to simultaneously optimize the investment and operation decisions under
uncertainty over the entire project horizon.
¾ How many FPSO/TLP
¾ Capacity of facility
¾ Time to build
¾ How many wells to drill
¾ Sub-sea/TLP well
¾ Production rate
Uncertainty:
¾ Sand quality (PI)
¾ Size of reservoirs
¾ Breakthrough time
Modeling*
Optimization
Process Synthesis/Design*
Process Operations
Process Control
min Z = f ( x, y )
s.t. h( x, y ) = 0
g ( x, y ) ≤ 0
x ∈ R n , y ∈ {0,1}m
MINLP- Mixed-integer Nonlinear Programming
Process Synthesis
Production Planning
Material flow
Information flow End
Demand for
Demands for
A, B & C
C
w(t+k|t)
y(t) N
LP, MILP, NLP, MINLP, Optimal Control t-1 t t+1 ... t+k ... t+N
Major contribution:
new problem representations, models & solution strategies
Plenary Lecture, ESCAPE-17, Bucharest, Romania, 27-30 May 2007 54
Contributions by Chemical Engineers
to Mathematical Programming
Global optimization
α-Branch and Bound
Spatial branch and bound methods
Parametric programming
Y jk
Disjunction
OR operator ∨ g ( x ) ≤ 0 , k ∈ K
Constraints
j∈J jk
k
c =γ Fixed Charges
k jk