Modeling at different Length and Time Scales Microscopic Mesoscopic Macroscopic Overall Rxn Rates 10 2 s Conc. Continuum Profiles Molec. Models Struct. 1 s Transport Models Statistical Mechanics Rate TST Consts. 10 -6 s Atomistic Elect. Struct. Simulation Thermo. Props. Molec. Dyn. 10 -12 s Quantum Mechanics DFT 10 -15 s 10 μ m 1 cm 1 A° 100 A° Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Paraffin Aromatization � Identify a catalyst formulation for light paraffin aromatization that is superior to Ga/H-ZSM-5 in terms of: � Higher Benzene, Toluene, Xylene (B/T/X) selectivity � Higher Hydrogen selectivity � Microkinetic model development for the kinetic description of the system Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Previous Work Microkinetic Analysis (Dumesic, 1993) First unified view of reaction engineering on catalytic surfaces Incomplete forward problem Lacks clear view of the design perspective (inverse problem) Empirical Catalyst Design (Baerns, 2000) First attempt to “design” catalysts Completely based on “guided” experiments – time consuming and expensive No fundamental understanding of the system (forward problem) Ertl, Rasmussen, Lauterbach: Surface Science Studies Steve Jaffe: Composition based modeling of large systems Jens Norskov: Computation based studies Symyx, Novodynamics, Lauterbach: Combinatorial HTE Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Reaction Network – Propane Aromatization on HZSM-5 = C y = C x C x Paraffin H + = Olefin C x adsorption/desorption adsorption/desorption = P x Poly-olefin =c Cyclic-olefin C x i n Active site H + i t i a C x t i o n Paraffin Aromatics C x H + n o i t P H 2 a Dehydro- n r o e C x-y g t o o l cyclization r y d s y i h s e D β -scission/Oligomerization = C x-y + C y + C x Hydride transfer • 31 gas phase species + 29 surface species + 271 reaction steps • Model with 31 ODEs, 29 algebraic equations • 13 parameters with up to 10 orders of magnitude bounds on each Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Ontological Informatics: Modeling Super Highway Data Advanced Parameter Chemistry Equation Statistical Optimizer Compiler Generator Analyzer Automatic generation of Solution of DAEs Sensitivity Analysis Reaction Description Performance differential algebraic Least squares Uncertainty Analysis Language Plus (RDL+) Curves equations (DAEs) Features Error Propagation Rate/Selectivity Chemistry Rules • DAE System: 30 to 1000 species and 10 to 50 parameters Reactions • Parameter search space is highly nonlinear with numerous local minima Grouping time GA based pseudo global parameter estimator Fast and efficient coverage of the parameter space • Optimization: Traditional least-squares and features-based objectives J.M. Caruthers, J.A. Lauterbach, K.T. Thomson, V. Venkatasubramanian, C.M. Snively, A. Bhan, S. Katare and G. Oskarsdottir, “Catalyst Design: Knowledge Extraction from High Throughput Experimentation”, In “Understanding Catalysis from a Fundamental Perspective: Past, Present, and Future”, A. Bell, M. Che and W.N. Delgass, Eds., Invited Paper for the 40th Anniversary Issue of the Journal of Catalysis , 2003. Katare, S., Caruthers, J. M., Delgass, W. N., Venkatasubramanian, V., “An Intelligent System for Reaction Kinetic Modeling and Catalyst Design”, Ind. Eng. Chem. Res. and Dev. , 2004. Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Model Refinement Procedure Infer new knowledge about models (mechanism) and data Model screening Model discrimination Model enhancement Crude models Initial data Enhanced models Richer data set Add/Delete models Planning new Model analysis experiments Data analysis Add/Delete data
Chemistry Rules for Propane Aromatization on HZSM-5 Representative Chemical Reactions Chemistry Rules 1. Alkane adsorption 2. Alkane desorption 3. Carbonium ion protolysis CH4 + 4. Carbonium ion dehydrogenation H2 + 5. Olefin adsorption 6. Olefin desorption 7. Aromatization 8. Beta-Scission + 9. Hydride Transfer + + 10. Oligomerization +
Modeling Super Highway: Reaction Modeling Suite (RMS) Reaction Network N M English Language Rules Reaction Description Language Plus A k 5 R k 6 Beta Scission k 1 Chemistry Label-site c1+ (find positive carbon) . . Label-site c2 (find neutral-carbon attached-to c1+) . D F Label-site c3 (find neutral-carbon attached-to c2) 8. Beta Scission k 3 H k Forbid (primary c3) transforms a carbenium ion into a k 2 4 Forbid (less-than (size-of reactant) 9) smaller carbenium ion and an olefin G E B Disconnect c2 c3) C . . . Increase-order-of (find bond connecting c1+ c2) k 7 Add-charge c3 Subtract-charge c1+ . . Beta Scission . Grouping Label-site c1+ (find positive carbon) Require (c1+ primary and product) Model Generator 8. a. Formation of a secondary carbenium ion set-k k1 Label-site c2+ (find positive carbon) is 20 times faster than a primary carbenium ion Mathematical Equations Require (c2+ secondary and product) b. Formation of a tertiary carbenium ion set-k 20*k1 = − dC / dt k C A 1 A Label-site c3+ (find positive carbon) is 60 times faster than a primary carbenium ion . = + − Require (c2+ tertiary and product) . / dC dt k C k D k B . B 1 A 4 5 set-k 60*k1 θ + θ + θ = 1 A B C . . . 100’s of DAE’s Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Results Data * Model C 2 H 6 C 3 H 6 Wt % C 3 H 8 Aromatics Space time x 10 4 [hr] Over prediction of C 2 s and under prediction of aromatics Venkat Venkatasubramanian, Plenary Lecture, * Lukyanov et. al., Ind. Eng. Chem. Res. , 34, 516-523, 1995 Foundations of Computer-Aided Process Design, Princeton, July 2004.
Results – Model Refinement Original Model Data Refined Model C 3 H 6 C 2 H 6 Wt % C 3 H 8 Aromatics Space time x 10 4 (hr) Refined model adds alkylation step to convert lighter alkanes to higher ones Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
GA Based Pseudo Global Parameter Estimator Performance comparison on test problems – worst case: 3 ODEs/5 parameters S.no. Name Esposito & Floudas (2000) GA based procedure Global optimizer υ υ Objective CPU s Objective Time Scaled CPU reported taken s (CPU s) min Time taken max *671/211 (% Saving) First order 5.0035 2.92 5.0122 1.18584x10 -6 9.25418x10 -6 0.76 (74) 1 irreversible chain 0.24 1.0000 20.05 0.9976 reaction 4.0000 3.9813 First order 2.0000 568.44 1.9759 1.8897x10 -7 8.1313x10 -6 1.43 (100) 2a reversible chain 0.45 40.013 6164.3 39.787 reaction 20.007 19.887 4.021 4.001 Same as 2a but 2.052 272.91 2.027 1.586x10 -3 1.589x10 -3 2b 0.5 1.59 (99) 39.45 39.22 with error in data 1899.82 19.62 19.50 Catalytic 12.214 12.246 79.77 2.65567x10 -3 2.68017x10 -3 3 0.38 1.21 (98) cracking of gas 7.9798 7.9614 1185 oil 2.2216 2.2351 4.5704x10 -6 4.5815x10 -6 Bellman’s 36.16 11.26 (69) 4 22.03094 22.2885 3.54 2.7845x10 -4 2.7899x10 -4 problem 12222 5.1981 5.2212 1361.8 Methanol-to- 1.2112 1.2320 19125 1.6363x10 -31 5 hydrocarbon 0 0.10652 0.10586 2.08 6.61 (100) 4.0059x10 -12 process 0 0 0.004665 Lotka-Volterra 3.2434 367.26 3.1434 1.24924x10 -3 2.39784x10 -3 6 0.56 1.78 (100) Problem 0.9209 9689.67 0.9583 Our zeolite model 31 ODEs, 29 Algebraic equations Reasonable comparison 2 hours on a Sun 400 MHz 13 parameters to experimental data solaris machine 9 concentration curves
Ontological Informatics for Catalyst Design: Summary � Represents reaction High Throughput High Throughput Chem istry Chem istry chemistries explicitly Experim ental Experim ental Rules Rules Data Data � Customizable for different catalyst chemistries Reaction Reaction Kinetic Modeling Kinetic Modeling Know ledge Know ledge Toolkit Toolkit Base Base ( Reaction ( Reaction � Results Modeling Suites) Modeling Suites) � Real-time evaluation of 100s of reaction pathways and 1000s of Correctness and Consistency Checking Correctness and Consistency Checking DAEs Reaction Reaction Network Network Application Application Application Generator Generator � Saved months of model Ontology Ontology Molecule Molecule Proposed Proposed Reaction Reaction development effort Classification Classification Mechanism Mechanism Classification Classification Reaction Reaction Reaction Molecule Molecule Molecule Reaction Reaction Reaction � Improved mechanistic Mechanism Mechanism Mechanism Ontology Ontology Ontology Ontology Ontology Ontology Chemistry Chemistry Ontology Ontology Ontology Knowledge Knowledge Chemistry Ontology Chemistry Ontology understanding and first Base Base Graphical User Interface Graphical User Interface principles models Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Proposed Framework for Knowledge Extraction from HTE – Kinetic Modeling New HTE for Model Discrimination via Experimentation Experiments in new composition regimes Measure critical variables identified High Throughput Experiments FTIR GC . . . Performance Curves Reaction Modeling Feature Reaction Modeling Feature Suite Extractor/ Rate/Selectivity Suite Extractor/ Automatic processing of Analyzer Automatic processing of Analyzer data data time Chemistry Rules Chemistry Rules Reactions Reactions Lumping Lumping Model Refinement Rules � Features mapping Suggestions for location of new rules
Catalyst Design Challenge Target Catalyst Rate/Selectivity Catalyst Library Compare Performance HTE time Target Catalyst Performance Model Revision Statistical Analyzer Rate/Selectivity k 1 Statistics/Neural-Nets Pseudo Global Optimizer A + S A-S Physical Model Feature Extractor k 2 C + R D Pseudo English {k} A-S +D k 3 Rule Compiler F time Kinetics Catalyst Performance AI/Systems Tools Catalyst Hybrid Model Reaction Modeling Suite Forward Model Recombination Selection Genetic Algorithms Inverse Model
LUBRIZOL: Fuel Additive Design EPA requirement: Minimize � intake-valve deposits (IVD) Approach: Fuel Additives � Performance measure � BMW Test for IVD � Stipulated to be less than 100 mg � over a 10,000 mile road test Expensive and time-consuming � testing Around $8000 for a single datum � Problem: Design fuel-additives � that meet desired I VD performance levels Intake Valve and Manifold Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Fuel-Additives : A Functional Description Additive Component that binds to deposit “precursors” Additive Component that keeps it in solution Fuel + Oil (Liquid + Vapor Mixture) Fuel-Additive Molecules High-Boiling Deposit Forming Precursors
First- -Principles Model for Additive Degradation Principles Model for Additive Degradation First Breakage of this bond Breaking of these bonds control “length” removes “dirt” carrying capacity totally Component “length” directly indicative of stability of additive Chemical nature of this component (polar/non-polar) controls “dirt” removing capacity The first-principle model tracks the structural distribution of fuel-additive with time due to reactive degradation
General Framework of Predictor Models General Framework of Predictor Models Ab Initio Methods Group Contribution Regression Structural Descriptors Secondary Performance Material/Molecule Primary Model Model Formulation Neural Networks Pattern Recognition Topological Molecular Techniques Mechanics
Modeling Degradation Modeling Degradation � Dynamic in nature � Modeled as first-order irreversible reactions for additive degradation � Rate constants in proportion to rates of pure thermal degradation � Distribution of molecular species differing in structure � Identified by effective "length" � Population balance model tracks distribution with time
From Stability to IVD From Stability to IVD � Define " amount of active additive " The fraction of the additive species that remains active (i.e. intact viable structure) in the fuel at any point of time � A dynamic quantity decreasing with time � Depends on additive distribution � Solubility correlated to IVD via regression � Linear models � Neural-Network models � Input descriptors are the amounts of active additive at different times
Forward Problem Approaches Data-driven/empirical models Empirically speaking, a polynomial expansion fits my data............ Fundamental Physics/Chemistry Expert Rules/ Heuristics ∂ ρ ( v ) + • ∇ ρ = − ∇ + ∇ • τ v ( v ) P ( ) ∂ If A goes higher, B should go higher ..... t ∂ ρ = − ∇ • ρ If C goes higher, D should go lower ..... ( v ) ∂ t How to integrate diverse scientific knowledge/information into a single unified knowledge architecture ?
Modeling Philosophy � All models are wrong…. … but some are useful! -- George Box (U. Wisconsin) � Hybrid Models � Combine First-principles and Data-driven models Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.
Hybrid Model for IVD Prediction Hybrid Model for IVD Prediction Amount of Active Additive Intake-Valve Deposit Additive A Additive B τ τ 2 τ τ 0 i N Time
Computer- -Aided Design of Fuel Aided Design of Fuel- -Additives Additives Computer Statistical/Neural-Net Correlation Physical Model Hybrid Model Intake Valve Deposit Additive Structure Additive Fuel Property Performance Engine Conditions Recombination Selection . . Genetic Algorithms
Previous Approaches to Inverse Problem � Previous Methodologies � Random Search � Heuristic Enumeration � Math Programming � Knowledge-Based Systems � Graph Reconstruction � Disadvantages � Combinatorial Complexity � Nonlinear Search Spaces � Local Minima Traps � Difficulties in Knowledge Acquisition � Difficulties in using high-level chemical/bio-chemical knowledge
Overview of Genetic Algorithms Overview of Genetic Algorithms � Definition Genetic Algorithms are stochastic, evolutionary search procedures based on Darwinian model of natural selection � Essential Components � Genetic Operators � Crossover � Mutation � Reproductive Plan � Fitness Proportionate Selection
Genetic Algorithms for Product Design Genetic Algorithms for Product Design � Global Search � Diversity of solutions � High potential for novelty � Global Optima � Development is de-coupled from forward problem � Robust to non-linearity � Population based search � Ability to provide several near-optimal solutions � Captures transparently the rich chemistry of the design problem
Overview of Genetic Algorithms Overview of Genetic Algorithms Initial Population Molecular Designs Calculate Fitness Select Parent(s) Select Operator Crossover Mutation Apply Operator Create New Population
Genetic Algorithms (GA) Genetic Algorithms (GA) GAs are stochastic evolutionary search procedures based on the Darwinian model of natural selection “ Survival of the fittest ” Fitness Calculn, Operators New I nitial Population Parent Selectn Population (random) Evolution
Genetic Operators: Crossover Genetic Operators: Crossover Single-point Crossover � Parent # 2 Parent # 1 H H CH 3 ] n [ ] n [ C O C C O C C O C O O O O H H CH 3 Offspring # 1 Offspring # 2 CH 3 H H ] n [ O ] n [ O C C C O C C O C O O CH 3 O H H
Mutation Operators Mutation Operators Main-chain and Side-chain Mutation � Parent: Offspring: H H H H H Mainchain Mutation C C C C C H H H H H Replace CH 2 n n by Offspring: Parent: H H H H Sidechain Mutation C C C C F H H Replace F by n n
Fitness Function for GA-based CAMD 2 Fitness (F) = exp - α P i – P P i,max – P i,min α = 0.0001 α = 0.0005 Pi – P Fractional Error (FE) = P α = 0.001 α = 0.01
Polymer Design Case Study • Base Groups 17 Main-chain Groups 15 Side-chain Groups -SO 2 - >C< -S- -O- -NH- -C2H5 -nC H -iC H -CH -H 3 3 7 3 7 O O O O O -F -Br -OH -Cl tC 4H O- -C- - -C- -C- O-C- 9 O -O -C- O O O O O -CN - CH - -C- O -C- CH - -OCH 3 3 -C- NH- -O-C- NH- -NH -C- NH- 3 X X X • Target Properties • Density • Glass Transition Temperature • Specific Heat Capacity • Thermal Expansion Coefficient • Bulk Modulus
Target Properties Glass Thermal Heat Bulk Density Target Polymer Transition Expansion Capacity Modulus (gm/cm 3 ) (K) J/kg K K N/m 3 1/K H H C C TP1: O C C O 5.18x10 9 O O H H 1.34 2.96x10 -4 1152.67 350.75 n H F H H TP2: 1.18 2.81x10 -4 2.51x10 9 225.24 1377.82 C C C C CH n H F H 3 CH 3 TP3: 2.90x10 -4 1135.10 5.40x10 9 O C O C 1.21 420.83 O CH n 3 CH 3 1.19 2.90x10 -4 5.39x10 9 TP4: 406.83 1073.96 S O C O CH n 3 TP5: 5.31x10 9 2.89x10 -4 O 995.95 SO 2 O 1.28 472.00 n TP6: 1.25 2.90x10 -4 421.12 6.12x10 9 O O C 1016.55 O n H H H H H H TP7: 3.85x10 9 2.98x10 -4 1455.90 1.06 322.55 C C C C C C N H H H H H O n
Polymer Design Case Study: Results Random MC, SC Random MC, SC Target Polymer Feasibility Constraints H H 12% 60% C C O C C O 184 240 282 213 O O H H n 36% 48% H F H H 411 522 C C C C 6 6 n H F H CH 3 CH 3 0% 12% - 193 O C O C 163 74 O CH 3 n 0% 0% CH 3 - - S O C O 861 589 CH 3 n 32% 48% 400 232 O SO O 2 175 142 n 8% 32% C O O 548 632 199 168 n O H H H H H H 100% 100% 61 64 C C C C C C N 217 198 H H H H H n O H H 68% 88% 210 109 O C C O C C 162 161 O O H H n 8% 4% CH CH 3 3 382 868 O O 144 70 CH 3 n Legend: Success Rate; Average Conv. Generation; Number with fitness >0.99
Near Optimal Designs Polymer Design Overall Error Fitness Target Polymer: #4 C H 3 0% 1.0 S O C O C H n 3 Case 1: Standard GA H H O C O C C 0.74% 0.995 H H n C H 5 O 2 O O O C O C 1.18% 0.991 C N n Case 2: Modified GA, Stability O H H O C C O C O 0.25% 0.999 H C H 3 n H C O 2 5 C S C O 0.991 1.10% H n Case 3: Modified GA, Stability & Complexity H O 0.21% 0.999 C O C H n C H O O 3 0.83% 0.995 C O C O C C H 3 n
Advantages of GAs for Product Design � Global Search � Diversity of solutions � High potential for novelty � Global Optima � Development is de-coupled from forward problem � Robust to non-linearity � Population based search � Ability to provide several near-optimal solutions � Captures transparently the rich chemistry of the design problem
Drawbacks of GAs Powerful generic method with some drawbacks • No convergence guarantees • Performance sensitive to parameters • Performance dependent on search space structure
Evolutionary Design of Fuel- -Additives Additives Evolutionary Design of Fuel Start Objective IVD desired < IVD limit ; Initial Population Maximize Solubility • Random • Knowledge Based Calculate Fitness (F) New Generation Structure ---> Hybrid Model Elitist Policy Fitness <---- Additive Performance Apply Operator Retain k best in population Select Parent(s) Head Linker Fitness Proportionate Branch Tail Termination Criteria • Last Generation ? No • Fitness = 1.0 ? Select Operator Crossover Mutation Yes Probabilistic Stop
Genetic Operators: Branch Crossover Genetic Operators: Branch Crossover Branch A L N K Site 2 J Parent-2 Parent-1 Branch Crossover N J K Offspring-2 Branch A moved to site-2 to L ensure feasibility of offspring-1 Offspring-1
Evolutionary Design of Fuel Additives: Results Evolutionary Design of Fuel Additives: Results Objective: Determine a structure with IVD < 10 mg Objective: Determine a structure with IVD < 10 mg Dosage: 50 PTB; Population Size: 25; Generations: 25 Dosage: 50 PTB; Population Size: 25; Generations: 25 Predicted IVD Run Rank/Identifier Fitness Structural Description (PLS-NN Model) 11.4 mg 0.997 Novel Structure. Infrequently used linker. 1, I-1 Novel Structure. Same tails as best 11.5 mg 0.996 2, I-2 structure, different heads and linkers I Variant of structure found in the BMW 12.0 mg 0.993 database. Same head and linkers, different 6, I-6 tails Novel Structure. Different from I-1 . 0.999 10.1 mg 1, II-1 Infrequently used linker component. Slight variant of additive structure found II 12.6 mg 0.989 in BMW and HONDA databases. 2, II-2 Different tails but same head and linker Minor variation of structure II-2 above. 0.983 13.2 mg 4, II-4 Slight modification of the head Novel Structure. Different from I-1 and 8.9 mg 1.00 1, III-1 II-1 . Commonly used components Variant of III-1 . One linker and tail 11.9 mg 0.994 2, III-2 III modified. Variant of structure II-2 above. Slight 0.993 12.1 mg modification of head. A linker and tail 3, III-3 inserted.
Rubber Parts in Service Tires, Treads, Hoses, Shock Absorbers, O- rings, Gaskets, Mounts …….. School of Chemical Engineering, Purdue University
Problem Definition Mixing (~ hrs) Curing (~ hrs) Thermal History Activator Accelerator Retarder Given the Desired Property/Performance of Part Rubber Interest, what is the Optimal Rubber Formulation? Rubber Formulation Filler Antidegradants Sulfur Oils Deformational Environment Thermal Environment Activator Accelerator Retarder Which materials to include? How much of them to include? Final Rubber Application (~ yrs) Filler Rubber Vulcanized Part Formulation What should be the processing conditions? Antidegradants Sulfur Oils Chemical Environment School of Chemical Engineering, Purdue University
What are Sulfur-Vulcanized Elastomers? Vulcanization S S S S + S 8 Activator Accelerator Retarder Rubber Accelerated Sulfur Vulcanization Filler Rubber Formulation Antidegradants Sulfur Oils School of Chemical Engineering, Purdue University
OBJECTIVE OF THIS WORK • Understand the Mechanistic Details of Accelerated-Sulfur- Vulcanization. N ZnO/Stearic-Acid C S N O S Activator Retarder Accelerator Benzothiazole Sulfenamide Rubber Filler NR Rubber Formulation Antidegradants Sulfur Oils S 8 • Provide a Mathematical Description of the Kinetics of Accelerated- Sulfur Vulcanization consistent with Mechanistic Chemistry AIChE Annual Meeting, Reno 2001 School of Chemical Engineering, Purdue University
COMPLEXITY OF VULCANIZATION REACTIONS W.Scheele (1956) 1 Perhaps no-where in chemistry is there encountered a field which even in its literature alone shows so many uncertainties and (possibly only apparent) contradictions as that of the vulcanization of rubber. 1 W. Scheele, O. Lorenz and W. Dummer, Rubber Chemistry and Technology , 29, 1 (1956) L.Bateman (1963) 2 Whilst it has long been appreciated, albeit intuitively, that sulfur vulcanization is a very complex chemical process, the actual complexity as revealed in the studies described above is probably far in excess of what has ever been envisaged. 2 L. Bateman, C.G. Moore, M. Porter and B. Saville, “The Chemistry and Physics of Rubber-Like Substances” , L Bateman, Ed., Maclaren & Sons Ltd., London, 1963, pp 449. P.J. Nieuwenhuizen (1997) 3 It must be considered remarkable that despite all the efforts and progress in the field of vulcanization during the past decade, one has to conclude that the statements of Scheele and Bateman, made 30 to 40 years ago, are still true to a great extent. 3 P.J. Nieuwenhuizen, Rubber Reviews, Rubber Chemistry and Technology , 29, 70 (1997) AIChE Annual Meeting, Reno 2001 School of Chemical Engineering, Purdue University
SCHEMATIC OF A VULCANIZATION CURVE Temporal Evolution of Crosslinks Concentration of Crosslinks, ν Induction Phase Curing Phase A S S S S S Reversion Phase S 0 S S Time Time Scale ~ Hrs Time Scale ~ Years AIChE Annual Meeting, Reno 2001 School of Chemical Engineering, Purdue University
CAMD Sub-Problems FORWARD PROBLEM Hybrid Models (Given Structure, Predict Properties/Performance) Neural Network Fundamental Models Properties/ Performance Material Structure/ ( P 1 , P 2 , ……..P N ) Formulation Property INVERSE PROBLEM Structure Space (Given Properties, Predict Structure/Formulation) School of Chemical Engineering, Purdue University
S Torque Cure Curve S S k 1 x + S ⎯ → ⎯ ⎯ A A S x + 8 8 THE OVERALL FORWARD MODEL Zn S A x + R ⎯ → ⎯ k 2 ⎯ B x S S N k 3 * ⎯ → ⎯ ⎯ S B B x C x Time S etc . Quantum Polymer Network Kinetics Mechanical Response Chemistry Over 1000 rubber parts in New Material failure critical functions Molecules Formulation Overcure Mixing + Manufac- Heat Transfer turing + Process Mold Filling Undercure Part Manufacturing Part Sub Shape Assembly Design Full Design Product Design & Operation ⎡ ⎧ ⎫ ⎤ ∂ ψ ∂ ψ ⎭ I − ∂ ψ ∂ ψ ∂ C C − 1 ⎥ • T = 2 ρ R ⎨ + I 1 ⎬ C + I 3 ˜ ⎢ • ∂ I 1 ∂ I 2 ∂ I 2 ∂ I 3 ∂ C ⎩ ⎣ ⎦ Constitutive Model Design & Spring-Dashpot Operation of Approximation Engine Complex with Chemical 3D Finite Sub-Assembly Aging Element Part Model School of Chemical Engineering, Purdue University
Schematic of a Vulcanization Curve N (ZnO + Stearic Acid) + Accelerator Activator C S N O S Active Accelerator Complex 2-morpholinothiobenzothiazole S 8 N N Active Sulfurating Agent A x C S Sx S C S S C Rubber Bound Intermediate B x C N (Crosslink Precursor) CH Sx S C S C C Rubber Bound persulfenyl • B x Sx* CH radical C C Vu x Polysulfidic crosslinks CH Sx C C Degradation Desulfuration C Aged Network School of Chemical Engineering, Purdue University
Accelerator Chemistry N N Bt = Accelerator + Activator C S N O C S S L Active Accelerator Complex N N N N S 8 C S Sx S C S x C C S Zn S Active Sulfurating Agent S S S S L Without Activator With Activator BtS-NR 2 BtSH + R 2 NH BtS-NR 2 + BtSH BtSSBt + R 2 NH BtS-SBt + S 8 BtS-S 8 -SBt BtS-SBt + S 8 BtS-S-SBt + S 7 BtS-S 2 -SBt + S 6 ••• BtS-S x -SBt + BtS-S y -SBt BtS-S z -SBt + BtS-S w -SBt A x = BtS-S x -SBt School of Chemical Engineering, Purdue University
Crosslinking Chemistry (I) CROSSLINKING CHEMISTRY (I) Active Sulfurating Agent C Rubber N N C N C S Sx S C CH Sx S C S S S Rubber Bound Intermediate (Crosslink Precursor) A x B x CH 3 HC C CH C H CH CH S S x S S N S C C S S x N + N S S C H S N C S S (R.34) N C SH S School of Chemical Engineering, Purdue University
820 Reactions, 107 Species + CTP -R 2 NH BtSNR 2 CDB + Pthalimide BtSH (Retarder Action) BtSS x SBt + BtS-SBt S 7 BtS-SBt … S 8 S 8 BtSS x SBt BtSS y SBt BtSS x SBt BtSSSBt BtS-SBt BtSS 8 SBt + BtSS y SBt Sulfurating Species BtS-ZnS x -SBt Sulfurating Species BtS-S x -SBt, BtS-Zn-S x -SBt BtS-S x -SBt, BtS-Zn-S x -SBt Sulfurating Species) (Regeneration of S 8 S 8 Rubber, RH BtS • (Scorch Delay) Crosslink Precursor ZnO Crosslink Precursor + Bt-SH BtSZnSBt RS x –SBt, RS x -Zn-SBt RS x –SBt, RS x -Zn-SBt Rubber, RH S 8 • RS-S y+8 Persulfenyl Radical S 8 Persulfenyl Radical • • • RS-S y • + Bt-S z Bt S z+8 RS-S y S 7 RS-S y+1 S 8 • S 8 S 7 • Rubber, RH • Bt S z+1 Bt S z+2 Crosslinks Crosslinks RSS y -R RSS y -R Rubber, RH Mechanisms (Degradation) (Desulfuration) Lumped S 8 Pickup RS-S y Sequential S 8 Pickup BtSZnSBt Cyclic Sulfide Main- Chain Modification, Both RSS y-1 R + BtSZnSSBt Conjugated Dienes/Trienes, Inactive thiols School of Chemical Engineering, Purdue University
POPULATION BALANCE EQUATIONS Equations for MBTS and other sulfurating species 14 d [ ] ∑ = − − − − A k [MBS] k [MBT] [MBS] k [A ] (r 1) A k [A ] [S ] 0 MBS MBS - MBT A - A 0 r A - S 0 8 dt = r 2 16 16 ∑ ∑ − − ∗ + ∗ − + 2 k [A ] k [A ] B 0 .5 k [E ] k [A ] Vu r A - R 0 A - BST 0 E - E 0 DESULF 0 r = = r 1 r 2 { } { } ⎡ ⎤ + + + + + + 2 2 [A ] 2 [A ][A ] 2 [A ][A ] 0 . 5 [A ] 2 [A ][A ] [A ][A ] 1 1 2 1 3 2 1 4 2 3 ⎢ ⎥ { } { } + + + + + + ⎢ 2 ⎥ 2 [A ][A ] [A ][A ] 0 . 5 [A ] 2 [A ][A ] [A ][A ] [A ][A ] 1 5 2 4 3 1 6 2 5 3 4 { } ⎢ ⎥ + + + + 2 2 [A ][A ] [A ][A ] [A ][A ] 0.5[A ] ⎢ ⎥ 1 7 2 6 3 5 4 { } ⎢ ⎥ + + + + 2 [A ][A ] [A ][A ] [A ][A ] [A ][ A ] ⎢ 1 8 2 7 3 6 4 5 ⎥ { } + + + + + ⎢ 2 ⎥ k 2 [A ][A ] [A ][A ] [A ][A ] [A ][ A ] 0 . 5 [ A ] A - A 1 9 2 8 3 7 4 6 5 ⎢ ⎥ { } + + + + + 2 [A ][A ] [A ][A ] [A ][A ] [A ][ A ] [ A ][ A ] ⎢ ⎥ 1 10 2 9 3 8 4 7 5 6 { } ⎢ ⎥ + + + + + + 2 2 [A ][A ] [A ][A ] [A ][A ] [A ][ A ] [A ][ A ] 0 . 5 [ A ] ⎢ 1 11 2 10 3 9 4 8 5 7 6 ⎥ { } + + + + + + ⎢ ⎥ 2 [A ][A ] [A ][A ] [A ][A ] [A ][ A ] [A ][ A ] [ A ][ A ] 1 12 2 11 3 10 4 9 5 8 6 7 { } ⎢ ⎥ + + + + + + 2 2 [A ][A ] [A ][A ] [A ][A ] [A ][ A ] [A ][ A ] [ A ][ A ] 0 . 5 [ A ] ⎣ ⎦ 1 13 2 12 3 11 4 10 5 9 6 8 7 14 16 d ∑ ∑ = − − + + [A ] 2k [A ] k [A ] [S ] 2k [A ] A k [A ] Vu 1 A - R 1 A - S 1 8 A - A 0 r DESULF 0 r dt = = r 2 r 2 13 ∑ − + ∗ ∗ 2k [A ] A k [E ] [E ] A - A 1 r E - E 0 1 = r 1 School of Chemical Engineering, Purdue University
POPULATION BALANCE EQUATIONS Equations for MBTS and other sulfurating species ⎡ ⎤ 14 [ ] [ ][ ] d ∑ = − − − − A k MBT MBS k [A ] [S ] k [A ] ⎢ (i 1) A ⎥ k [A ] 0 A2 A3 0 8 A4 0 i C1 0 ⎣ ⎦ dt = i 2 ⎡ ⎤ ⎡ ⎤ 16 16 ∑ ∑ − − k [A ] ⎢ B * ⎥ k [ A ] ⎢ Vu ⎥ C4 0 i R 1 0 i ⎣ ⎦ ⎣ ⎦ = = i 1 i 1 ⎡ ⎤ 16 [ ] d ∑ = − − − + A 2 k [A ] k [A ] [ S ] 2 k [ MBTS ] [ A ] k [A ] ⎢ Vu ⎥ 1 C1 1 A3 1 8 A 4 1 R1 0 i ⎣ ⎦ dt = i 1 ≤ ≤ ≤ ≤ ⎧ ⎧ [ S ], 2 i 6 0 , 2 i 6 [ ] d = − − − + + 8 ⎨ ⎨ A 2 k [A ] k [ A ] (i 1) k [A ] [ A ] k [ A ] > > i C1 i A 3 i A4 0 i A 3 i ⎩ ⎩ dt 0 , i 7 [ S ] i 7 8 Equations for Crosslink Precursors d [ ] = + B ( k [MBTS ] k [E *] ) 0 C1 C7 1 dt [ ] d = − + − − + B k [ A ][ B *] 2 k [A ] ( i 1 ) k [ B ] k [E *] + i A4 0 i C1 i C 2 i C7 i 1 dt School of Chemical Engineering, Purdue University
RATE-CONSTANT DETERMINATION FACTS Total of 820 different reactions 107 Coupled Ordinary Nonlinear Differential Equations 9 Optimizable Rate Constants 2 ⎡ ⎤ exp ν − ν q m ( k ) [ ] ∑∑ ⎢ i , j ⎥ i , j = min , k k , k , k ...... k 1 2 3 P ⎢ ⎥ ν exp k = = ⎣ ⎦ i 1 j 1 i , j s . t . m time points, q concentrations [ ] d ν = φ ν ( , c , k ) dt ≥ k 0 = k Rate Constants exp ν = Experiment al Data Point at time i and conc j i , j School of Chemical Engineering, Purdue University
Example Writing Population Balance Equations • B 2 E • Potential Breaking Sites C N k A S* C C + S2* S CH C C N S C HC S S S B 2 C N k B * C C + S S S S* CH B • • E 2 ⎛− ⎞ ⎛− ⎞ [ ] ) [ ] d ( E E = − + = = ⎜ ⎟ ⎜ ⎟ 0 0 A B B k k B , k k exp , k k exp 2 A B 2 A A B B ⎝ ⎠ ⎝ ⎠ dt RT RT [ ] [ ] [ ] [ ] d d • • = = B k B , E k B 2 A 2 A 2 dt dt [ ] [ ] [ ] [ ] d d • • = = B k B , E k B B 2 2 B 2 dt dt School of Chemical Engineering, Purdue University
Population Balance Equations Equations for MBTS and other sulfurating species [ ] 14 8 d ∑ ∑ = − − − A k [MBT] [MBS] k [A ] (r 1) A k [A ] [S ] x = 0 0 MBS- MBT A - A 0 r A - S 0 y dt = = r 2 y 1 16 16 ∑ 1 ∑ − − ∗ + ∗ − 2 k [A ] k [A ] B k [E ] k [A ] Vu r A - R 0 A - BST 0 E - E 0 DESULF 0 r 2 = = r 1 r 2 − 13 13 x ∑∑ + k A A A - A x y = = x 1 y 1 8 14 16 d ∑ ∑ ∑ = − − + + [A ] 2k [A ] k ( [A ] - [A ] ) [S ] 2k [A ] A k [A ] Vu x = 1 1 A - R 1 A - S 1 0 y A - A 0 r DESULF 0 r dt = = = y 1 r 2 r 2 13 ∑ − + ∗ ∗ 2 k [A ] A k [E ] [E ] A - A 1 r E - E 0 1 = r 1 8 14 d ∑ ∑ = − − + [A ] 2k [A ] k ( [A ] - [A ] ) [S ] 2k [A ] A 2 ≤ x ≤ 13 x A - R x A - S x x - 1 y A - A 0 r dt = = + y 1 r x 1 − 14 x ∑ − − − + ∗ ∗ (x 1) k [A ][A ] 2 k [A ] A k [E ] [E ] A - A 0 x A - A x r E - E 0 x = r 1 x - 1 ∑ + k [A ][ A ] A - A r x - r = r 1 School of Chemical Engineering, Purdue University
Final Set of Population Balance Equations ⎡ ⎤ ⎡ ⎤ . . ⎢ ⎥ ⎢ ⎥ . . ⎥ ⎢ ⎥ ⎢ ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ ( ) ⎢ Ax ⎥ ⎢ f k , k , k , k , Ax ⎥ 1 01 02 1 4 ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ ( ) ∗ ⎢ ⎥ ⎢ ⎥ Bx f k , k , k , Ax , Bx , Bx 2 1 2 4 ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ d ⎢ ⎥ ⎢ ⎥ Initial Conditions: A 0 (0) = [MBS] 0 S 8 (0) = [Sulfur] 0 . . ⎢ ⎥ dt ⎢ ⎥ ⎢ . ⎥ ⎢ . ⎥ = ( ) ⎢ ⎥ ⎢ ⎥ ∗ ∗ A x(>0) (0) = B x (0) = B *x (0) = Vu x (0) = MBT(0) = E(0) = 0 Bx f k , k , k , Ax , Bx , Bx ⎢ ⎥ ⎢ 3 2 3 4 ⎥ ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ . . ( ) ⎢ ⎥ ⎢ ⎥ ∗ Vu f k , k , Bx , Vu ⎢ ⎥ ⎢ ⎥ x 4 3 6 x ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ . ⎢ . ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ . . ⎢ ⎥ ⎢ ⎥ ( ) ⎢ ⎥ ⎢ , , ⎥ S f k A S 8 5 01 0 8 ( ) ⎢ ⎥ ⎢ ⎥ MBT f k , Ax ⎢ ⎥ ⎢ ⎥ ( 6 1 ) ⎥ ∗ ⎢ ⎥ ⎢ ⎣ ⎦ ⎣ ⎦ E f k , k , k , k , Ax , Bx , Bx , E 7 2 4 51 52 Most simple kinetic model that explains all the important features of sulfur vulcanization. Total of 820 different reactions considered with 107 coupled ordinary nonlinear differential equations and 9 optimizable rate-constants. School of Chemical Engineering, Purdue University
Population Balance Model - Predictions Effect of Accelerator and Sulfur T = 330 F Legend T = 330 F (1.0, 4.0) (0.5, 4.0) + (1.0, 2.0) x (0.75, 2.0) (0.5, 2.0) Open Symbols = Experimental Data Open Symbols = Experimental Data Solid Line = Model Predictions Solid Line = Model Predictions School of Chemical Engineering, Purdue University
Population Balance Model – Different Temperatures T = 298 F T = 310 F T = 298 F T = 310 F Open Symbols = Experimental Data Open Symbols = Experimental Data Solid Line = Model Predictions Solid Line = Model Predictions School of Chemical Engineering, Purdue University
THE CATERPILLAR GRAND CHALLENGE S S S To q r u e Cure Curve k 1 x + S ⎯ → ⎯ ⎯ A A x + 8 S 8 Zn S k 2 A x + R ⎯ → ⎯ ⎯ B x S S N k 3 * ⎯ → ⎯ ⎯ B x B x S C Time S . etc Quantum Polymer Network Kinetics Mechanical Response Chemistry Over 1000 rubber parts in New Material failure critical functions Molecules Formulation Overcure Mixing + Manufac- Heat T ransfer turing + Process Mold Filling Undercure Part Manufacturing Part Sub Shape Assembly Design Full Design Product Design & Operation ⎡ ⎧ ⎫ ⎤ ∂ ψ ∂ ψ I − ∂ ψ ∂ ψ ∂ C C − 1 • T = 2 ρ R ˜ ⎨ + I 1 ⎬ C + I 3 ⎢ ⎥ • ⎩ ∂ I 1 ∂ I 2 ⎭ ∂ I 2 ∂ I 3 ∂ C ⎣ ⎦ Constitutive Model Design & Spring-Dashpot Operation of Approximation Engine Complex with Chemical 3D Finite Sub-Assembly Aging Element Part Model School of Chemical Engineering, Purdue University
Spatial Profile of the State-of-Cure for the Thermal History in the Mold e l i f o r Actual Part Design P l a m r e h T > 99 97-99 e l i f 95-97 o r 93-95 P e 91-93 r u 89-91 C 87-89 85-87 School of Chemical Engineering, Purdue University
Result Summary Vulcanization Vulcanization Chemistry Mechanistic Mathematical Insights into Evaluation Quantification Actual Part Design • Developed Population Balance • Incorporation of Population • Evaluated all possible Model based on mechanistic Balance model with a finite mechanisms critically for details to describe the different element code to predict accelerated sulfur aspects of cure. spatial cure profile. vulcanization • Single set of parameters that • Identification of undercured • Identified the simplest set of predicts cure details for all and overcured regions reactions that describe all the formulations at all temperatures. visually. important aspects of cure. • Model extended for retarders, filled systems and other accelerator and elastomer classes. School of Chemical Engineering, Purdue University
S S S To q r u e Cure Curve k 1 x + S ⎯ → ⎯ ⎯ A A x + 8 S 8 THE OVERALL FORWARD MODEL Zn S k 2 A x + R ⎯ → ⎯ ⎯ B x S S N k 3 * ⎯ → ⎯ ⎯ B x B x S C Time S . etc Quantum Polymer Network Kinetics Mechanical Response Chemistry Over 1000 rubber parts in New Material failure critical functions Molecules Formulation Overcure Mixing + Manufac- Heat T ransfer turing + Process Mold Filling Undercure Part Manufacturing Part Sub Shape Assembly Design Full Design Product Design & Operation ⎡ ⎧ ⎫ ⎤ ∂ ψ ∂ ψ I − ∂ ψ ∂ ψ ∂ C C − 1 • T = 2 ρ R ˜ ⎨ + I 1 ⎬ C + I 3 ⎢ ⎥ • ⎩ ∂ I 1 ∂ I 2 ⎭ ∂ I 2 ∂ I 3 ∂ C ⎣ ⎦ Constitutive Model Design & Spring-Dashpot Operation of Approximation Engine Complex with Chemical 3D Finite Sub-Assembly Aging Element Part Model School of Chemical Engineering, Purdue University
CAMD Framework Forward Problem Prediction Prediction Rubber Accelerator Cure-State Sulfur P 1 P 2 P 3 ... P n P 1 P 2 P 3 ... P n Modulus Activator Stress-Strain Filler Structure or Formulation Structure or Formulation Product Performance Stiffness Retarder Design Design Inverse Problem Inverse Problem T he solution of INVERSE PROBLEM involves searching for the optimal rubber formulation that has the desired macroscopic performance. School of Chemical Engineering, Purdue University
Genetic Algorithms (GA) GAs are stochastic evolutionary search procedures based on the Darwinian model of natural selection ⎡ ⎤ ⎛ ⎞ 2 − P P ∑ ⎢ ⎜ ⎟ ⎥ = − α i i, des Fitness (F) exp ⎜ ⎟ − ⎢ ⎥ P P ⎝ ⎠ ⎣ ⎦ Formulation = [1 4 0.1 30 330] i, max i, min Fitness I nitial Population Fitness Calculn, (random) Parent Selectn School of Chemical Engineering, Purdue University
Genetic Algorithms (GA, contd) “ Survival of the fittest ” Fitness Fitness Calculn, Operators New Parent Selectn Population Evolution School of Chemical Engineering, Purdue University
Formulation Representation Accelerator Accelerator Accelerator Activator Activator Activator Retarder Retarder Retarder Filler Filler Filler Rubber Formulation Rubber Formulation Rubber Rubber Rubber Antidegradants Antidegradants Antidegradants Sulfur Sulfur Sulfur Sulfur Oils Oils Oils Binary Representation Phr of Different Ingredients 1 0 1 0 1 1 0 0 • • • 1 School of Chemical Engineering, Purdue University
Genetic Operators Crossover Parent 1 Parent 2 1 1 0 0 1 1 0 0 0 0 1 1 1 1 1 1 0 0 1 1 1 1 0 0 1 1 0 0 1 1 1 1 0 0 0 0 0 0 0 0 Offspring 1 Offspring 2 1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 0 1 0 1 0 1 0 1 1 1 1 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 Mutation Fitness 1 1 0 0 1 1 0 0 0 0 1 1 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 1 1 0 0 1 1 0 0 1 1 Formulation School of Chemical Engineering, Purdue University
Inverse Problem (Results) • Binary Representation • Fitness Proportionate Selection • Crossover Probability = 0.8, Mutation Probability = 0.2 • Population Size = 40, Number of Generations = 40, Elitism = 10% Example 2 Desired Property Optimal Formulations Formulation Fitness T max Vu max σ 100 σ 200 T max (time to reach max cure) = 16 min [1.75 1.25 0.05 0 315] 0.9999 15 69.92 0.91 1.63 Vu max (crosslink @ T max ) = 71.5 mol/m 3 [0.50 4.00 0.10 0 310] 0.9999 16 71.57 0.92 1.65 Stress (100% Elongation) = 0.92 MPa [1.75 1.50 0.05 0 310] 0.9996 17 75.75 0.97 1.75 [0.75 2.75 0.10 0 315] 0.9994 13 67.88 0.88 1.58 Stress (200% Elongation) = 1.65 MPa [3.00 0.75 0.10 0 315] 0.9991 17.5 65.34 0.85 1.52 [2.75 1.00 0.25 0 330] 0.9990 10.5 70.46 0.93 1.67 School of Chemical Engineering, Purdue University
BEST THREE FORMULATIONS Formulation Fitness T max Vu max σ 100 σ 200 [1.75 1.25 0.05 0 315] 0.9999 15 69.92 0.91 1.63 [0.50 4.00 0.10 0 310] 0.9999 16 71.57 0.92 1.65 [1.75 1.50 0.05 0 310] 0.9996 17 75.75 0.97 1.75 New Formulations [0.75 2.75 0.10 0 315] 0.9994 13 67.88 0.88 1.58 80 [3.00 0.75 0.10 0 315] 0.9991 17.5 65.34 0.85 1.52 Concentration of Crosslinks, mol/m 3 70 [2.75 1.00 0.25 0 330] Solution Found in 0.9990 10.5 70.46 0.93 1.67 24th generation. 60 3 i.e.only 960 out 50 1 of a total of 131072 40 2 formulations were searched 30 ~ 0.73 % 20 10 0 0 10 20 30 40 50 60 Time, min School of Chemical Engineering, Purdue University
Interactive Software Used At Caterpillar on a Daily Basis School of Chemical Engineering, Purdue University
Pharmaceutical Product Development and Engineering Raw Chemicals Drug Synthesis Drug is converted into Particles Formulation Preliminary product recipe • Two Products Process Development Drugs Preliminary process recipe Documents Scale/up -Tech Transfer Adjusted process recipe Manufacturing A Delayed Product 10/ 11/ 2005 1
• Prescription drug recalls • 176 in 1998 • 248 in 2001 • 354 in 2002 • Schering-Plough Corp. recalled 59 million asthma inhalers in 1999 and 2000 -- unknown number were shipped empty. • Semiconductor Manufacture 6 σ • Chemicals 5 to 5.5 σ • Pharma 2.5 σ LIPS Venkat Venkatasubramanian, Keynote Lecture, European Congress in Chem. Engg, Copenhagen, Sept 2007 2
Pharma Industry’s Scale-up Challenge engineering pharmaceutical LIPS Venkat Venkatasubramanian, Keynote Lecture, European Congress in Chem. Engg, Copenhagen, Sept 2007 3
Purdue Ontology for Pharmaceutical Engineering ( POPE ) s c i t s M i r a u t h e h e m d a n t a i c s a e l e m r t o d n e o l i s s i c e D ModLAB Information OntoMODEL 4
Ontologies Developed So Far in Our Project � Equipment ontology � Standards referenced: STEP, AP231, FIATECH � General recipe ontology � RecipeElement, UnitProcedure, Operation etc. � Standards referenced: ISA S88, S95, OntoCAPE � Process safety ontology � Deviation, Cause, Consequence etc. � Material ontology for pharmaceutical product development � Flow property, Angle_of_Fall, Carrs_Index etc. � Reaction mechanism ontology � Molecule, Atom, Bond, Reaction etc. � Referenced: Chemistry Development Kit � Model ontology � Guideline ontology � Referenced: GuideLine Interchange Format LIPS Venkat Venkatasubramanian, Keynote Lecture, European Congress in Chem. Engg, Copenhagen, Sept 2007 5
Material Property Ontology in Protégé LIPS Venkat Venkatasubramanian, PSWC, April 2007, Amsterdam 6
Experiment Ontology Experimental Methods Details Experiment Data Files LIPS Venkat Venkatasubramanian, PSWC, April 2007, Amsterdam 7
Guideline Ontology for Dosage Form Formulation Follows GLIF (Guideline Interchange Format), A standard ontology for clinical guidelines LIPS Venkat Venkatasubramanian, Keynote Lecture, European Congress in Chem. Engg, Copenhagen, Sept 2007 8
9 Model Ontology
Recommend
More recommend
