Computer-Aided Methods & Tools for the Pharmaceutical Industry - Rafiqul Gani Department of Chemical Engineering

Page created by Allan Vaughn
 
CONTINUE READING
Computer-Aided Methods & Tools for the Pharmaceutical Industry - Rafiqul Gani Department of Chemical Engineering
C APEC

 Computer-Aided Methods &
 Tools for the Pharmaceutical
            Industry
                Rafiqul Gani
     Department of Chemical Engineering
      Technical University of Denmark
         DK-2800 Lyngby, Denmark
            www.capec.kt.dtu.dk
              rag@kt.dtu.dk
Statements from Industry
 We develop and provide customized computer-aided solutions in
 process/product development, planning and operations to assist our
 company (pharmaceutical and specialty chemical products) to
 introduce new products to the market faster with lower initial cost of
 goods while improving cost effectiveness of their development and
 manufacturing groups - CAPEC member company

 One of the key challenges facing pharmaceutical companies is to
 reduce the time to market and cost of goods of their products whilst
 continuing to comply and exceed stringent regulatory conditions -
 CAPEC member company

 Need to make design decisions in the early stages of the process
 (design) lifecycle so that aspects of energy conservation, environment,
 waste, etc., can be incorporated - CAPEC member company, …

 Our (CAPEC) role is to provide the methods & tools so that
           the above can be accomplished!
         Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   2
The Challenge

                                                    The Product Tree
                                                       The objective
                                                     (challenge) is to
                                                        identify the
                                                     important fruits
                                                      (products), the
                                                      optimal path to
                                                      reach them, the
                                                     feasibility of the
                                                        process, …..
       Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)    3
Dimension of the problem: Number of isomers
Cn    CnH2n+1OH               CnH2n+2                 CnH2n                  CnH2n-2
      p-alcohol               Alkanes                 Alkenes                Alkynes
1     1                       1                       0                      0
4     2                       2                       3                      2
8     39                      18                      66                     32
12    1238                    355                     2281                   989
16    48865                   10359                   93650                  38422
20    2156010                 366319                  4224993                1678969
 How many of these compounds are known? Can we afford
 to ignore this knowledge when developing new products
 and/or new processes?
           Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)             4
Need for models                                                                C APEC

                                 Information                               Problem
                                 (knowledge,                                Design
  Process/                          data)
  Product
                                                                           Production
 Operation                         Simulation
                                                                           Planning
Environment
                                       Model                                Analysis
 Business

  System
         Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                5
Research Programs in CAPEC                                                                 C APEC

                         Tools Integration E
                                                                          G
 E           C Synthesis/Design/Analysis

                                                                Safety & Hazards
 Pollution & Waste

                           Process Models B
        Product

                                                                    Process
                      Property/Phenomena
                            Models    A                                            Numerical Tools
                     B     Product Models                                               Databases F
                     Control & Operation                    D

                         Process Integration E

                      Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)           6
Activities
                                  Activitiesin
                                             inCAPE
                                                CAPEProblems
                                                     Problems

                                                         Parameter fitting
  Choosing among                                                                        Simulation
                               Thermodynamic                 for the
 Process Alternatives                                                               Design Calculation
                               Model Selection           Thermodynamic
   Feasibility Study                                                                  Optimization
                                                              Model

   Properties of Equilibrium
unknown molecules Analysis       Model Selection Guide
                                                          Regression Regression Process        Properties
                                                             Input     Result
                                       TMS                                     Conditions       Values
                               Thermodynamic Model
                                     Selection
  Propred                                                 TML
                                          Thermodynamics Model Library
    3.5

 Pure component      Pure properties
                     Export / Import
    properties
   predeiciton                              CAPEC Thermophysical Properties Database
                                            - Pure component properties and parameters
                                            - Mixture properties and parameters
                                            - Experimetal data

                  Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                  7
Correlation & Prediction of Properties
! Group contribution approach for pure component
  and mixture properties
! Applications in specialty chemicals, fine chemicals,
  pharmaceuticals
! Electrolyte systems modeled
! Maintenance of library modules & model parameter
  tables
! CAPEC database of pure component and mixture
  properties (including solvents)
! Integrated computer aided methods & tools
I&EC Research (2002), Fluid Phase Equilibria (2001), J Chem Eng Data
  (2001), Trans IChemE (2000), I&EC Research (1999)

            Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   8
Property Estimation                                                                        C APEC

Step 1:                                                     System conditions

Problem analysis                        Compounds present
                                                                                  Operation type
                                          in the system

Step 2:
Generation/creation of                                  Generation of problem
                                                           specific model
                                                                                             Property Model
                                                                                                 Library
property model

Step 3:                                                 Parameter estimation

Parameter Estimation
                                                        Validation of generated         Property calculation
Step 4:                                                          model                    (eg., simulator)

Validation of Property
Model

            Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                            9
New Group Contribution Method                                                          C APEC
      Property (X)        Left-hand side of Eq. 1               Right-hand side of Eq. 1
                              [Function f(X)]                (Group-Contribution Terms)
 Normal Melting Point          exp(Tm/Tm0)                Σ iN iTm1i+ Σ jM jTm2j+ Σ kO kTm3k
 (Tm)
 Normal Boiling Point           exp(Tb/Tb0)               Σ iN iTb1i+ Σ jM jTb2j+ Σ kO kTb3k
 (Tb)
 Critical Temperature           exp(Tc/Tc0)                Σ iN iTc1i+ Σ jM jTc2j+ Σ kO kTc3k
 (Tc)
 Critical Pressure (Pc)       (Pc-Pc1) -0.5 -Pc2          Σ iN iPc1i+ Σ jM jPc2j+ Σ kO kPc3k
 Critical Volume (Vc)              Vc-Vc0                 Σ iN iVc1i+ Σ jM jVc2j+ Σ kO kVc3k
 Standard Gibbs               Gf-Gf0           Σ iN iGf1i+ Σ jM jGf2j+ Σ kO kGf3k
 Energy at 298 K (Gf)
 Standard Enthalpy of         Hf-Hf0           Σ iN iHf1i+ Σ jM jHf2j+ Σ kO kHf3k
 Formation at 298 K
 (Hf)
 Standard Enthalpy of         Hv-Hv0                  Σ iN iHv1i+ Σ jM jHv2j
 Large    increase
 Vaporization at 298 K in application range; Improved accuracy;
 (Hv)
 Automatic       group selection
 Standard Enthalpy of
                                   from theΣ imolecular
                            Hfus-Hfus0
                                                                structure
                                              N iHfus1i+ Σ jM jHfus2j+ Σ kO kHfus3k
 Fusion (Hfus)
                        Fluid Phase Equilibria (2001)
              Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                   10
Application Example                                                      C APEC

  Analyze the solubility of Fentanyl
    * Generate data (using ProPred)
    * Insert generated data into database
    * Perform solid-liquid phase diagrams

       Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)      11
Separation Boundaries: Distillation

 VLE system                                           VLLE system
   Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   12
Separation Boundaries: Crystallization

    Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   13
Process & Product Modeling                                                   C APEC

• Modeling of micro-organisms
• Modeling of fermentation                            Models are the basis for all
  processes                                           further work. Once the
• Modeling of hybrid distillation                     process/operation model
  operations and other non-                           has been established, it is
  conventional separation                             analyzed, solved, combined,
  processes
                                                      manipulated, etc., in other
• Modeling of fixed-bed reactor                       projects for development of
  systems
                                                      algorithms for synthesis,
• Special purpose models (eg.,
  uptake of pesticides, drugs, …)                     design, control; for better
• Computer aided process                              understanding of process &
  modeling toolbox                                    product; for inventory, ….
 C&CE (1999, 2001, 2002), CAPE-NET report, Modeling Workshop (1999), Escape-
                   10,11 12, PSE-2000, Trans IChemE (2000)
            Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)      14
What is computer aided modeling?

    Problem
    definition                                                        Human
                       System
                    characteristics
                                               Problem
                                                 data
                                                                        Model
                                                                     construction
                                                Model
                                               solution
                         Model
      Model            verification
  calibration &
                                                                      Computer
    validation
   Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)             15
Model Construction Steps
                               M O D ELS

                            M A T H E M A T IC A L
                                 M O D ELS

                          PRO CESS M O D ELS

           • Derive the model equations
            • Analyze model equations
• Translate the model equations to a solvable form;
 create library for use with a simulator or for on-line
                        solution
     A computer aided system assists the user in
           performing the above tasks

    Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   16
Model construction, generation & reuse,
                                                                  Extract equations
                      Decomposition                               from library
                                                                       * Balance
                                                                       Equations
                                                  Mathematical
                                                     model                *Constraint
                                                                           Equations
                                                                          *Constitutive
                                            Building block                 Equations
                          Aggregation

Define Boundary           Describe System                   Identify Building
                                                                  Block

        Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)            17
Application of the modeling tool-box
                                                                 Calculator or
                                                                    solver

                              Modeling                                Model
 Process or                   Tool-box                              library in
  Model                       (MoT in                               simulator
                               ICAS)

                             Connect                              Create own
                           model object                           customized
                            to external                            simulator
                           applications

        Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)          18
Construction of an operation & design model
                       F3                         Operation/Design Model
  F1
                                                  1. Charge Feed (open F1
                                F4                   & close F2)
         Reactor
                                                  2. Close F1
                                                  3. Heat until
                                 F2                  temperature = 340 K
  Reaction : A → B                                4. Control temperature
                                                     at 340 K
  Maximum conversion                              5. Charge solvent by
  of 50% A at T = 340 K                              opening F3
  Extract B from reactor                          6. Extract B by opening
  with a solvent!                                    F4
  Solvent ID and effects                          7. …….
  need to be modeled
         Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   19
Process/Product Synthesis & Design
  Generate new process/product alternatives                     Determine
               Process ?                              * Optimal process/product
 Raw                              Products?
Materials                                             * Process/product/operation
    Generate better alternatives for an               alternatives
        existing process/product                      * Design of control system
                                                      * Operating conditions
                                   purge
                                                        How? Make decisions based
                                    sepa-               on the available knowledge
       mixer        reactor         rator
                                                         (data) and/or calculations
 Raw                      Products                         using the appropriate
Materials                                                     methods & tools
                Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   20
How do we use models in process design?
                                              Design
                                            parameters              Synthesis &
                Simulation                                            Design
 Raw                                          Results
material                                                         Iterative (trial &
             Process Model                                        error) approach
                                             Product
  T, P, X,
compounds                               Properties

           Constitutive Model

           Forward Problem: Service Role for
             Process & Phenomena Model
             Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)      21
Current Approaches: Limited Search Space
                                                                Constitutive Model
                                                 Constitutive Model

                                Constitutive Model
                                   * Choice of a constitutive model
                                  implicitly defines the search space
                                   * Do we know enough to derive a
                                            single model?
                                 * Can we solve simulation and/or
                                optimization problems with multiple
                                constitutive models representing the
                                           same variable?
       Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)       22
More efficient method of solution
                                            Design
                                          parameters              Synthesis &
              Simulation                                            Design
 Raw                                        Results
material                                                      Iterative approach
           Process Model
                                           Product
  T, P, X,
compounds                            Properties
                                                               Design targets &
       Constitutive Model                                         feasibility

    Forward Problem: Service & Advice Role
           for Constitutive Model(s)
           Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)        23
Integrated Process and Product Design
                                          Molecular Design

                                          Discrete Decisions
                                        (e.g. type of compound,
    Given set of molecular            number of functional groups)                   REQUIRED
    groups to be screened
       (building blocks)
                                                                                   A framework for
                                         Continuous Decisions
                                        (e.g. operating conditions)               tracking properties

                      Designed components                       Constraints on property values
                   (e.g. raw materials, MSA's)                  obtained by targeting optimum
                                                                     process performance

                                           Process Design

                                           Discrete Decisions
                                      (e.g. structural modifications)                Desired process
    INTERFACE
                                                                                       performance
Constitutive equations,                  Continuous Decisions                  (e.g. recovery, yield, cost)
                                        (e.g. operating conditions)
e.g. property relations
                Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                               24
Problem Definition: Reverse Problems

  Reverse Simulation Problem: Given, Inputs
    & Outputs, calculate property clusters
               (design targets)

Reverse Property Calculation Problem: Given,
Property (design targets), determine molecules,
     mixtures and/or conditions (T, P, x)

       Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   25
C APEC
Other examples of reverse problem
    Consider the reverse problem of simulation – given the
    design variables, solve the process model equations to
        determine the corresponding property values

     Mass Exchange Operation: Given flow, mass of solute
      to be recovered, determine needed property (design)
                   target value for solubility
      For given solubility, generate list of solvents and the
      operating temperature that satisfy the desired target
      Note that in the 1st step, calculation is composition
    independent & 2nd step is reverse of property prediction

          Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)      26
An example of product design
.

Structure                                                            Can we find better
                                                                        solvents than
                                                                      benzene, toluene
                                                                      or cyclohexane?
                                                                     Can we “design”
                                                                       drugs/chemical
Name           Morphine                                                 products with
Tm (K)
       ½
               528                                                        desirable
δ (MPa)        26.3
                                                                         properties?
Figure 1: Molecular structure and properties of morphine

    Define the product needs and then identify the molecules that
                        match these needs!
                 Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)      27
CAMD Framework
                                                                             Start
• Solutions of
  CAMD problems                                           No suitable
                                                        solutions (due to

  are iterative.        Candidate
                         selection
                                                         performance,
                                                          economic or
                                                        safety concerns)
                                                                              Problem
                     (final verification/
• Problem                comparison
                       using rigorous
                                                                            formulation
                                                                              (identify the
                                                                              goals of the
  formulation          simulation and
                        experimental             Promising                  design operation)
                                               candidates have
  controls the          procedures)             been identified

  success.                                               Result                                 Method and
                                                      analysis and                               constraint
• Essential qualities Finish                           verification
                                                         (analyse the
                                                                                                 selection
                                                                                                 (specify the
  vs. (un)desirable                                       suggested
                                                          compounds
                                                                                                design criteria
                                                                                                based on the

  qualities.                                            using external
                                                            tools)
                                                                                                   problem
                                                                                                 formulation)

• Connectivity with                                                            CAMD
  external tools,                                                             Solution
                                                                               (identify
  data and methods.                                                           compound
                                                                              having the
                                                                               desired
                                                                              properties)

                Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                          28
Example: Design of solvent for Phenol
                                     SLE Diagram for Phenol/Solvent system
       320
    Benzene
 • Structure   is used as Methyl sec-Butyl for
                          a solvent        Ether dissolving
                                                       2,2-Dimethyl-1-propanal
                                                              Phenol.
    Because of environmental considerations, it is desirable
    to300replace Benzene.
                 Benzene (experimental Tfusion )
           280
 • Specifications:
   T (K)

      – Should dissolve Phenol at ambient temperature conditions at
       260
          least
   Solvation     as well as Benzene.
             energy                   -3.863 kcal/mol                     -7.081 kcal/mol
                                                      Methyl sec-Butyl Ether (estimated T fusion )
      – Have physical properties similar to Benzene.
       240        Benzene (estimated Tfusion )
      –   A  stripper    column is used to regenerate Water used for
          equipment cleaning between batches.
      –220Not
           0
                pose the0.2  same environmental0.4
                                                            problems
                                                              0.6
                                                                             as benzene.
                                                                                   0.8             1
                                                   x(Phenol)

                 Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                       29
Computer Aided Molecular (Product) Design
                                         Pre-design                                          Design (Start)

                                    Interpretation to   A set of building blocks:          A collection of group
       "I want acyclic             input/constraints
                                                        CH3, CH2, CH, C, OH,                   vectors like:
     alcohols, ketones,                                 CH3CO, CH2CO, CHO,                 3 CH3, 1 CH2, 1 CH,
   aldehydes and ethers                                  CH3O, CH2O, CH-O                        1 CH2O
   with solvent properties                                          +
    similar to Benzene"                                    A set of numerical                All group vectors
                                                              constraints                   satisfy constraints

                                           Design (Higher levels)                              Start of Post-design
  2.order                                                           Refined property                  CH2       CH2         CH3
                   CH2       CH2          CH3
   group     CH3         O         CH                             estimation. Ability to        CH3         O         CH

                                                                   estimate additional                                CH3
                                   CH3
                                                                    properties or use
                                                                                                                CH3
                             CH3                                  alternative methods.
Group from                                                                                            CH2       CH          CH3
                   CH2       CH           CH3
other GCA    CH3         O         CH2
                                                                  Rescreening against           CH3         O         CH2
  method                                                             constraints.

       Feasible alternatives found but do we have a sustainable
                      process to manufacture it?
                    Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                                               30
Synthesis & Design of Distillation Trains
                                                                                                                             A   B

                                       Secondary Separation Efficiencies at P=1 atm.

       0.3

                                                                                                                         A
                                                                                                                         B
      0.25
                                                                                                                         C
                                                                                                                         D
                                                                                                                         E
       0.2

                                                                                                     Propane iButane
                                                                                                     iButane nButane
                                                                                                                                 C
SSE

      0.15
                                                                                                     iPentane nButane
                                                                                                     nPentane iPentane

       0.1
                                                                                                                                 D

      0.05

        0
             0   0.1   0.2       0.3       0.4        0.5       0.6        0.7       0.8   0.9   1
                         Liquid composition, mole fractions of the lightest com pounds

                                                                                                                                 E

                  Order the driving force diagrams in terms of
                 fij| max; configure the distillation train in terms
                   of fij| max; design each distillation column in
                           terms intersection on DyDx line.
                                                 Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                     31
Visualization of Synthesis & Design
- System : H2O - (l)Asparagine -
          Alanine - Serine
- Products : (l)Asparagine,
          Alanine, Serine
- Solubility description:
          Solubility product-
          (l)Asparagine, Alanine,                                     8
          Serine                                                   11
- Number of chemical species: 12                                               4
                                                               2
- Phase diagram type: quaternary                     1
- Thermodynamic model:
          Electrolyte NRTL
                                                         3                6                           10
 Simultaneous problem               1
                                        Mixer
                                                2   Cryst.     4
                                                                    Mixer
                                                                               7   Cryst.     8   Cryst.
                                                    373.15 K                       313.15 K       343.15 K
solution and visualization                          5                                   9             12     11
 for batch & continuous                             Alanine                         L-Asp          Serine

   operations/processes                             14
                                                                   Splitter
                                                                          13
Trans IChemE (2000), C&CE (2000)                                    Purge
Example: Flowsheet synthesis/design
      Flowsheet Synthesis
                       C                            2

     The operator based reduced                                         Feed 1
                                                                                                                      Mixer     Product
                              0.1    0.9
       models derived from the
        original mass balance
                          0.2                               0.8

      models are linear,
                      0.3    thereby                              0.7

      making them applicable to                                         Feed 2                        Splitter
                  0.4                                                   0.6
     simple graphical techniques
                              0.5                                             0.5

                                                                                                                    Byproduct
                        0.6                                                         0.4

                                                   Target
                  0.7                                                                     0.3

                                                 Feed 1
                                                                                                                 Visual Simulation
            0.8                                                                                 0.2
                                                                                             Composition free design
      0.9
                                                                     Feed 2
                                                                                  0.1          calculations, but the
                                                                                            compositions can be back
C3                                                                                    C
              0.1         0.2         0.3   0.4   0.5    0.6   0.7     0.8    0.9         calculated directly since the
                                                                                                            1

                                                                                                flowsheet satisfies
                                                                                              fundamental balances
                                    Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)              33
Examples : Blending/mixing Problem
 Composition free
 reduced model
 C1m = c1C11 + c2C12
 C2m = c1C21 + c2C22
 C3m = 1 - C1m + C2m

          Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)   34
PROBLEM DEFINITION                 TOOL BOXES

ICAS                            Flowsheet
                                Components / Reactions
                                                                Design / Synthesis      Analysis
ADD TO THE SYSTEM                                                Solvent/Fluid                 Energy
                                Units of Measure
    New Components
     (Property Prediction)
                                Constitutive Models              Equipment                  Environmental
                                What to Solve
                                Method of Solution               Flowsheet                     Control
    New Reactions
                                Set/Initialize Variables         Control
                                Output (Detail/Form)                                    Thermodynamic
  New Models
(Model Generation)
                                                              Parameter Estimation       Property
                                                                 Thermo-model            Phase Diagrams
    DATABANKS                    INFORMATION                                             Expert System
                                                                 Kinetic Model
                                   STORAGE

SIMULATOR
MANAGER
 Model Equations             Adaptation                 Analysis                     Solvers
  Balance Equations           Linearization                Degrees of Freedom         AE / ODE / DAE
  Constraint Equations        Reduction                    Index / Sparse Pattern     PDE
  Constitutive relations      Identification               Partitioning / Ordering    LP / NLP

                      Symposium on Science Collaborations, AstraZeneca,
                                                                  RHS2002 (R.Gani)    MILP / MINLP
                                                                                                 35
    RHS for the units that are solved together                     X
Industrial Collaboration: Examples
• Computer Aided Molecular Design - solvents,
  process fluids, product qualities, …
• Modeling & analysis of electrolyte systems
• Optimization of a crystallization process
• Modeling, simulation & control of complex
  distillation operations
• Synthesis of batch operations (process route
  selection)
• Properties of new products (CAPEC Database) &
  molecule structure generation
• Reaction system analysis - alternative reaction
  paths
• Solubility of complex (new) chemicals &
  downstream separation
 ……. Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani) 36
Process-Product Development in Industry
          Research route

                                          Generate route
                                             options                                   Outline f/s &
                                                                                          costs

                                                                                                                                  Product
                                                                                                                                specification
                                                                                                          Early
      Selection criteria                                                                                                           market
                                                                                                       formulation
                                                                                                                                  forecast
                                                                                                                                  business
                                                                                                                                   targets
    SHE impact
Market requirements
   (quality; tox)                         ROUTE SELECTION
                                                                                                                 Design, f/s,
      Activity                                                                                                  cost estimate
   VPC / margin
      Capital
                                                                         FF&P development
             THE KEY DECISION
                  POINT
                                                       Process                                 Ongoing
                                                     development                             development

      FF & P = formulation, fill                                                                                                   Decision to
             and pack                                                                                                                invest

                                                             Small                   Field trials;
                                                           manufacture               registration

                                            Research Results of KT
                           Symposium on Science Collaborations, AstraZeneca, 2002 (R.Gani)                                             37
CAPEC Member Companies                              C APEC

• Current membership: 22 companies (+ IMP, Mexico)
• Access to research reports, databases & software
• Specialized collaborative research projects & exchange
  visits/training
• Annual discussion meeting                                38
You can also read