Thermodynamic Modelling at NPL

1
Thermodynamic Modelling
Thermodynamic modelling at the National Physical
Laboratory
Hugh Davies
8 March 2002
2
Content

• National Physical Laboratory
• NPL Materials Centre
• Materials Processing Team
• Thermodynamics and Process Modelling Group
• MTDATA and the calculation of phase equilibria
• Models, Materials and Databases
• Tour of Modules
• Data assessment
• Conclusion

3
What We Are

• NPL is the UKsNational Standards Laboratory
• for physical measurements

4
Organisation - Scientific
Mechanical Acoustical
CMAM
NPL Materials centre
CLM
MATC
Length
CISE
CIRM
CS
Information Systems Engineering
Ionising Radiation
CEM
CBTM
Corporate Services
Electromagnetic
Basic Thermal
CTM
COAM
Optical Environmental
Time
5
Who are we?

• NPL Materials Centre (MATC) is one of the worlds
  leading materials research laboratories
• We are the largest scientific centre at the
  National Physical Laboratory in Teddington
• We employ over 90 scientists working on a wide
  range of materials measurement areas. Currently,
  work is divided into four areas
• Measurements for materials processing
  manufacturing
• Characterisation performance of materials
• Surface properties engineering
• Materials systems components, design, modelling
  and databases

6
(No Transcript)
7
Thermodynamic Modelling MTDATAAlan Dinsdale
Hugh Davies John Gisby Susan Martin Jim
Robinson
The prediction of phase equilibria using
thermodynamic data

• Wide range of applications eg
• Alloy Development, Materials processing
• Energy conversion, Joining
• Corrosion and deposition, Environmental control
• Extraction and recycling
• Electronic and magnetic materials

• Comprehensive set of databases eg oxides alloys
  aqueous, semiconductors .
• Robust algorithms for reliability
• Available for a wide range of computers
• Assessment module

8
Thermodynamic Modelling Engineering
ToolkitsAlan Dinsdale Hugh Davies John Gisby
Susan Martin Jim Robinson
The use of predictive materials chemistry for
process simulation and control

• Provides link to third party commercial packages
  eg CFx, Fluent, Physica, MAGMASoft
• Custom built engineering toolkits for specific
  industries eg lighting applications, casting of
  alloys

• MTDATA Application Programming Interface for user
  written software
• Api is self documenting for easy maintenance

9
Heat Flow Properties for Metals and OxidesRob
Brooks Lindsay Chapman Alan Dinsdale Peter
Quested Jim Robinson
Development of methods for the measurement and
prediction of thermal conductivity and enthalpy
for modelling.

• DSC
• Drop calorimetry
• Single pan calorimetry
• Laser flash
• Prediction of properties

• Al
• Fe,Cu,NI
• Ni
• Oxides

Virtual measurement system Software to predict
property values
Comparison of DSC, drop calorimeter and
calculated data
10
Thermodynamic modelling using MTDATA
MTDATA This software/data package is used to
predict the complex chemistry in multicomponent
multiphase systems, using critically assessed
thermodynamic data. It is used by companies and
universities worldwide for such diverse
applications as continuous casting of steel, lamp
chemistry, pollution control and
pyrometallurgical extraction.
11
What is MTDATA ?

• Designed to calculate phase/chemical equilibria
  with maximum ease and reliability
• Interactive
• User application programming
• Despite reports to the contrary MTDATA does not
  work like other comparable programs
• A true Gibbs energy minimisation
• A non-linear optimisation problem with linear
  constraints
• NPL Numerical Optimisation Software Library
• No initial guess required
• Mathematical guarantee that G reduces each time
  it is evaluated
• Result can be proven to be a solution to the
  posed problem
• Very high reliability

12
Typical graphical outputs

• Phase diagrams
• Thermodynamic properties
• Phase compositions
• Assessment monitoring

13
Example of convergence testing

• ISO 9001 release acceptance criterion is gt 99.99
  
• No initial guess
• Aggressive random testing for 12 system/model
  types
• 10000 equilibria in each routine test
• Up to 5000000 in additional tests
• Self testing facility is built into every MTDATA

14
Too good to be true ?

• What can go wrong - not much
• near 100 likelihood of convergence
• slow convergence could be a problem
• Miscibility gaps and systems with ordering
• in some dimension the Gibbs energy appears to
  have multiple minima
• heuristic methods give good reliability
• MTDATA is very strong on calculations in systems
  with miscibility gaps and phases whose Gibbs
  energy is not an explicit function of system
  composition

15
Models and Materials

• Data for a phase consists of
• Unary G as f(T,P) possibly magnetic with one or
  more sublattices
• Gex as f(T,P,x), typical models
• Redlich Kister
• Quasi chemical
• Modified Wagner dilute
• Modified Pitzer
• ternary or reciprocal interactions
• To model a wide range of mostly inorganic
  materials
• Alloys
• Semi-conductors
• Oxides
• Salts
• Aqueous solution
• Gas phase and stoichiometric solids
• Others (eg Polymer solutions, Organic vapours and
  Bio systems)

16
Databases availability (1)
17
Databases availability (2)
18
Databases availability (3)
19
Typical output from UNARY database
20
Typical output from P_UN database (1)
21
Typical output from P_UN database (2)
22
Typical output from TCFE database
23
Typical output from MTSOLDERS databaseAg-Cu-Pb-Sn
system
24
Typical output from MTSOL database
25
Typical output from MTAL database
26
Typical output from SGSUB database
27
Typical output from NPLOX2 database
28
MTDATA family structure
MTDATA for Windows v4.X
Core MTDATA
MTDATA for PC
MTDATA for UNIX
User application
Binding to 3rd party application
Application Support Environment
MTDATA for Windows v5.X
NPL Engineer's Tool Kits
NPL Virtual Measurement systems
29
NPL Virtual Measurement Systems

• Designed to be used by non specialists
• Simple interface
• Reliability of MTDATA
• Integration with MS Excel

30
NPL Engineer's Tool Kits

• Bespoke software for compact fluorescent lamp
  design
• Materials selection built in
• Industry specific outputs, integrated with MS
  Excel

31
Relative Light Output results
32
EHECATL General MTDATA/Fluent binding
33
Aqueous Chemistry(1) U-H2O system
34
Aqueous Chemistry(2) H2O-Am(NO3)3-EDTA
35
Ag to Pb0.5Zn0.5 isopleth including gas phase
36
Ag to Pb0.5Zn0.5 isopleth including gas phase
37
Ag to Pb0.5Zn0.5 P-x diagram at 800 K
38
LM25 equilibrium solidification
39
LM25 Scheil solidification (1)
40
LM25 Scheil solidification (2)
41
Metal metal halide phase equilibria
42
Modules - UNARY

• Property calculation for individual phases in a
  unary system

43
Modules - UNARY

• Property calculation for individual phases in a
  unary system

44
Modules - UNARY

• Property calculation for individual phases in a
  unary system

45
Modules - GPLOT

• Property calculation for individual phases in a
  binary system

46
Modules - GPLOT

• Property calculation for individual phases in a
  binary system

47
Modules - BINARY

• Very simple calculation of binary phase diagram
• No initial guesses
• 30 phases considered from NPL Alloy database

48
Modules - BINARY

• Very simple calculation of binary phase diagram
• No initial guesses
• 30 phases considered from NPL Alloy database

49
Modules - BINARY
50
Modules - TERNARY

• Very simple calculation of isothermal ternary or
  reciprocal phase diagrams
• No initial guesses
• Interactive labelling
• 24 phases considered from NPL Oxide database

51
Modules - TERNARY

• Very simple calculation of isothermal ternary or
  reciprocal phase diagrams
• No initial guesses
• Interactive labelling

52
Modules - TERNARY
53
Modules - MULTIPHASE

• Interactive interface to the core equilibrium
  algorithm
• Fixed T,P(V),x or variable in many ways
• Open systems
• Activity, composition, enthalpy constraints and
  Para equilibrium

54
Modules - MULTIPHASE
Cp of 9-component alloy
Adiabatic combustion of carbon
55
Modules - MULTIPHASE
56
Modules - APPLICATION

• A platform for user written application code
  embedded into MTDATA
• NPL provided built in applications
• Isopleths, T-P, P-x, liquidus projections, fixed
  T sections and solidification simulations

57
Modules - COPLOT

• Predominance area calculations
• Log Activity diagrams
• Pourbaix diagram with pH and EH on axes

58
Modules - THERMOTAB

• Thermodynamic functions for chemical reactions
• Automatic equation balancing
• Direct two way database interface

59
Data Assessment

• Completely general framework for data assessment
• MTDATA command language allows virtually any
  experimental information to be included
• No practical limit to the size of system being
  assessed
• Any thermodynamic model parameter can be included
  in the assessment process
• Real-time graphical monitoring of objective
  function and parameter values
• Extensive graphical results reporting
• Integration with the rest of MTDATA

60
Typical experimental data input

• Water activity data at 298.15
• Macro T/K molality act(H2O)
  uncert reference
• "actH2O.mac" 298.15 0.1 0.998004
  0.0001 "actH2O-29874PIT/MAY"
• "actH2O.mac" 298.15 0.2 0.99639
  0.0001 "actH2O-29874PIT/MAY"
• "actH2O.mac" 298.15 0.3 0.994889
  0.0001 "actH2O-29874PIT/MAY"
• "actH2O.mac" 298.15 0.4 0.993443
  0.0001 "actH2O-29874PIT/MAY"
• "actH2O.mac" 298.15 0.5 0.99202
  0.0001 "actH2O-29874PIT/MAY"
• NiSO4 enthalpy of formation at various
  concentrations (82WAG/EVA)
• Macro n(H2O) DfHsol uncert
  reference
• "DfHdil.mac" 200.0 -958505 500
  "DfH-29882WAG/EVA"
• "DfHdil.mac" 300.0 -958672 500
  "DfH-29882WAG/EVA"
• "DfHdil.mac" 400.0 -958797 500
  "DfH-29882WAG/EVA"
• "DfHdil.mac" 500.0 -958906 200
  "DfH-29882WAG/EVA"
• "DfHdil.mac" 800.0 -959169 200
  "DfH-29882WAG/EVA

61
Data Assessment monitoring

• Real time monitoring of objective function
• Display of changes in parameter values

62
Post Assessment Reports

• Residual scatter plot or bar chars
• Correlation plots
• Property plots
• Full Integration within MTDATA

63
Conclusions

• MTDATA is a very general tool for calculating
  phase and chemical equilibria with very high
  reliability and no need for initial guesses at
  the system/phase composition
• A wide range of models and databases allow
  application to a large number of different types
  of problem
• Complex user applications can be built using
  calls to MTDATA routines from users code or 3rd
  party applications
• Produced by NPL the UK National Standards
  Laboratory
• Thanks to Alan Dinsdale, John Gisby, Jim
  Robinson, Susan Martin and Malcolm Rand
• Funding in part from DTI EID and all commercial
  users of MTDATA

64
(No Transcript)