ASIM Workshop on Fundamentals of Modeling and Simulation...

Lehrstuhl für ThermodynamikProf. Dr.-Ing. H. Hasse

Molecular Modeling and Simulation in Process Engineering

Hans Hasse1, Jadran Vrabec2

ASIM Workshop on Fundamentals of Modeling and Simulation

1Lehrstuhl für Thermodynamik, TU Kaiserslautern2Lehrstuhl für Thermodynamik und Energietechnik, Universität Paderborn


Technology Vision 2020: The U.S. Chemical Industry

Chemical & Engineering Science

Manufacturing & Operations

Supply Chain Management

Information Systems

Engineering Computational Technologies

ComputationalMolecular Science

=> Link between Engineering and Chemistry=> New processes, products, materials

Process Modeling & Simulation

Operations Simulation & Optimization

ComputationalFluidDynamics

Fields of Major New Developments


(fs) 10-15

(ps) 10-12

(ns) 10-9

(μs) 10-6

(ms) 10-3

100

10-10 10-9 10-8 10-7 10-6 10-5 10-4

(nm) (μm)

Time / s

MesoscaleMethods

Continuum Methods

Semiempirical QM

Ab initioQM

MolecularForce Fields

Length / m

Molecular Modeling and Simulation Methods


Further progress from new simulation methods and software

Moore‘s Law

GFL

OPS

/ G

IPS

Year


Bridging Scales

QuantumSystems

Molecular Systems

ContinuumSystems

Born-Oppen-heimer MD

MD LargeSystems

Mesoscale Systems

HybridCFD

COSMORS

Examples


Molecular Methods in Chemical Process Industries

Current Applications @ Evonik Degussa:CatalysisThermo-physical dataPolymersCrystallizationParticle technology


COSMO-RS @ Evonik

Quantum chemistry based method for prediction of thermo-physical properties

Quantitative predictions

Quantum chemistry based predictionof energy of molecular interactions

Crude classical assumptions for entropy

Close co-operation between engineersand quantum chemists

Benchmark against phenomenolgicalgroup contribution methods (UNIFAC)


Force-Field Methods @ Evonik

Example: MD simulations of Water-Polymer interface

Qualitative results


Mesoscale Methods @ EvonikExample: Simulation of particle morphologies

α = 1

α = 4

α = 16

Semiquantitative results


Molecular methods are:already used especially attractive if no sufficiently accurate

- experiments - calculations with phenomenological methods

are possiblerecognized as a future key technology

Industrial Applications of Molecular Methods in Process Engineering


Molecular Modelling (Force Fields)

Geometry:Bond lengths and angles

Electrostatics:Position and strengthof dipoles, quadrupoles,partial charges

Dispersion and Repulsion:Parameters of Lennard-Jones potentials

Many parameters


Model Parameters from Quantum ChemistryGeometry

HF with small basis set (z.B. 6-31G) or DFT methodsElectrostatics from electronic density distribution

MP2 with small polarizable basis set (e.g., 6-311+G**)Molecule embedded in dielectric cavity for modeling dense fluid phase (COSMO)

Dispersion and RepulsionRequires simulation of arrangements of at least two moleculesCCSD(T) or MP2 with large basis sets (TZV or QZV)Very high computational effort Unsatisfactory accuracy

Fit to thermo-physical data preferred


Molecular Dynamics (MD)Numerical solution of Newtonian equations of motionDeterministicStatic and dynamic properties

Molecular Simulation Basic Methods

Monte-Carlo (MC)Statistical MethodEnergetic acceptance criteriaStatic properties only


Direct Simulation of Phase EquilibriaExample: Vapor-liquid equilibirum of Ethylene Oxide @ 375 K

3500 molecules


Phase Equilibrium from Grand Equilibrium Method

( ) ( ) ( )μ ≈ μ + ⋅ −l l li i 0 i 0p p v p p

Pseudo grand canonical simulation(Specification of V, T) ( )μ l

i p

Specs: T, x

Result: p, y

Liquid Vapor

Simulation:Chemical potentialsPartial molar volumes


10

20

30

40

Exe

cutio

n tim

e / 1

000

s

Number of processorsSX-8

XC6000Cacau

StriderMozart

1 2 4 8 16 32 64

High Performance Parallel Computing

Strider

SX 8

Cacau

XC6000

Own parallel FORTRAN codes: ms2 thermo-physical propertiesls1 nano-scale processes

Scaling on selected hardware platforms:


Polar Two Center Lennard-Jones ModelsQ

ε

L

σ

4 Model parametersε energyσ sizeL elongation

μ dipoleorQ quadrupole

dispersionrepulsion

polarity

μ /

Models of 80 simple pure componentsParametrization: VLE data only

Simulation Exp. (corr.)

typ. deviation < 2%

Vapor pressure


Pure Component Models: ExtrapolationsJoule-Thomson Inversion

p / M

Pa

T / Kred: critical data

Symbols: SimulationLinies: Reference EOS

Ethylene

Oxygen

Nitrogen


Ethylene Oxide

Worldwide annual production about 18 Mio. tonsUse: PET and anti-freezeProperties:

- explosive- toxic- highly flammable- cancerogenic- mutagenic

Explosion @ Sterigenics Intl., Ontario, CND (2004):4 wounded, hall destroyed


Industrial Property Simulation Challenge 2007

Industrial Fluid PropertiesSimulation Collective


Molecular Model of Ethylene Oxide

3 LJ Sites (one for the oxygen atom, one for each methylene group)1 static point dipole along symmetry axisRigid, non-polarizableAdjustment of five parameters (σO, εO, σCH2, εCH2, μ) to experimental VLE data


IFPSC Challenge 2007 : Problem DescriptionDevelopment of a new molecular model for Ethylene OxidePrediction of 17 properties in 3 categories

Benchmarked to “reference data”

Vapor liquid equilibria /thermal propertiesSaturated densitiesVapor pressureEnthalpy of vaporizationCritical propertiesNormal boiling temperatureSecond virial coefficient

Second derivatives/surface tensionHeat capacityIsothermal compressibilitySurface tension

Transport propertiesShear viscosityThermal conductivity


Deviations from Reference Data

uncertaintyof reference

Round-Robinmodel

new modelscore:331/350

deviation from experiment / %-40 -20 0 20 40 60

sat. vap. thermal conductivitysat. liq. thermal conductivity

sat. vapor shear viscositysat. liquid shear viscosity

surface tensionsat. vap. isoth. compressib.

sat. liq. isoth. compressib.sat. vapor isob. heat capacitysat. liquid isob. heat capacity

critical temperaturecritical density

normal boiling temperatureenthalpy of vaporization

vapor pressure2nd virial coefficient

sat. vapor densitysat. liquid density



















deviation from reference / %


IFPSC Party 2007


Ethanol

3 LJ sites plus 3 point chargesPoint charges model both electrostatics and H-bonding

δρ = 0,3 % δp = 3,7 %


H-Bonded-Species in Methanol + CO2Geometrical H-bonding criterion of Haughney et al.Equimolar mixture @ 350 K, 0.1 MPa

Legend:Donor: light blueAcceptor:single: orange double: red


1H-NMR Spectroscopy of H-Bonding Mixtures

ppm

293,15 K

338,15 K p = 15 MPa

Experiment

Simulation

Methanol - CO2


Overview Pure Component ModelsNon-polar, 1CLJ

Neon (Ne), Argon (Ar)Krypton (Kr), Xenon (Xe)Methan (CH4)

Dipolar, 2CLJD

Kohlenmonoxid (CO)R11 (CFCl3)R12 (CF2Cl2)R13 (CF3Cl)R13B1 (CBrF3)R22 (CHF2Cl)R23 (CHF3)R41 (CH3F)R123 (CHCl2-CF3)R124 (CHFCl-CF3)R125 (CHF2-CF3) R134a (CH2F-CF3)R141b (CH3-CFCl2)R142b (CH3-CF2Cl)R143a (CH3-CF3) R152a (CH3-CHF2)R40 (CH3Cl)R40B1 (CH3Br)CH3IR30B1 (CH2BrCl) R20 (CHCl3)

Dipolar, 2CLJD (contd.)

R20B3 (CHBr3)R21 (CHFCl2)R12B2 (CBr2F2) R12B1 (CBrClF2)R10B1 (CBrCl3)R161 (CH2F-CH3)R150a (CHCl2-CH3) (1,1,2) CHCl2-CH2ClR140a (CCl3-CH3)R130a (CH2Cl-CCl3)C2H5Br (CH2Br-CH3)(1,1) CHBr2-CH3CH2F-CCl3(2,2,2) CHClBr-CF3R112a (CCl3-CF2Cl) CHF=CH2CF2=CH2C2H3Cl (CHCl=CH2)CHCl=CF2 CFCl=CF2CFBr=CF2

Quadrupolar, 2CLJQ

Flour (F2)Chlor (Cl2)Brom (Br2)Iod (I2) Stickstoff (N2)Sauerstoff (O2)Kohlendioxid (CO2)Kohlendisulfid (CS2) Ethan (C2H6)Ethylen (C2H4)Ethin (C2H2)R116 (C2F6) C2F4C2Cl4Propadien (CH2=C=CH2)Propin (CH3-C≡CH) Propylen (CH3-CH=CH2)SF6R14 (CF4)R10 (CCl4)R113 (CFCl2-CF2Cl)R114 (CF2Cl-CF2Cl)R115 (CF3-CF2Cl) R134 (CHF2-CHF2) (1,2) CH2Br-CH2BrCBrF2-CBrF2CHCl=CCl2

Dipolar, 1CLJD

R32 (CH2F2) R30 (CH2Cl2)R30B2 (CH2Br2)CH2I2

Polar, Muti-CLJ

iso-Butan (C4H10)Cyclohexan (C6H12)Methanol (CH3OH)Ethanol (C2H5OH)Formaldehyd (CH2=O)Dimethylether (CH3-O-CH3)Aceton (C3H6O)Ammoniak (NH3)Methylamin (NH2-CH3)Dimethylamin (CH3-NH-CH3)R227ea (CF3-CHF-CF3)Schwefeldioxid (SO2)Ethylenoxid (C2H4O)Dimethylsulfid (CH3-S-CH3)Blausäure (NCH)Acetonitril (NC2H3)Thiophen (SC4H4)Nitromethan (NO2CH3)Phosgen (COCl2)Benzol (C6H6)Toluol (C7H8)Chlorbenzol (C6H5Cl)Dichlorbenzol (C6H4Cl2)Cyclohexanol (C6H11OH)Cyclohexanon (C6H10O)


Unlike interaction A-B:Electrostatics fully predictiveLennard-Jones parameters from combination rules

Molecular Modelling of Mixtures

( )AB A B+= /2σ σ σ

AB A B=ε ε εξ ⋅

A A

B B

σA, εA

σB, εB

σAB, εAB

or

Fit to one experimental data point p(T,x) oder H(T)

ξ = 1Predictions

ModifiedLorentz-Berthelot


Vapor-Liquid Equilibrium ofHeptafluoropropane + Ethanol

Simulation, =1 + ExperimentPeng-Robinson EOS ξ

InternationalFluidPropertiesSimulationChallenge2006

Data basis:=> VLE @ 283 KProblem:=> H-bonds


Simulation, ξ = 1□, ∆ Experiment Simulation, ξ fitted

InternationalFluidPropertiesSimulationChallenge2004

,

Henry’s law constant von Oxygen in Ethanol


Extrapolation to multicomponent mixtures

+ Experiment

PR-EOS, kij fitted to binary subsystems

Simulation, ξ fitted to binary subsystems

R14 + R23 + R13

Fully predictiveNo ternary parameters


MD Simulation of Nanoscale Processes: Condensation

N = 40 0001 CLJ


N = 40 0001 CLJ

MD Simulation of Nanoscale Processes: Condensation


EthaneSimulation Class. nucleation theoryLaaksonen et al.

Prediction of Nucleation Rates

Carbon DioxideSimulation Class. nucleation theory Laaksonen et al.


Molecular Simulation of Hydrogels

Examples for applications:Super-absorberContact lensesDrug DeliverySensorsActors (e.g., micro-valves)Biocatalysis

200 µm3 actors

flow channel

What are Hydrogels?Three-dimensional hydrophilic polymer networksExtreme swelling/shrinkingVery sensitive to surroundings & conditions


Swelling of HydrogelsParameters:

Temperature pH-valueSalt(s)Solvent(s)Co-polymersCrosslinker

Influence of temperature

Theta-temperature

PNiPAM, MBA

PNiPAM by electron-microscope


• Mainly PNiPAM (numerous experimental data)

MD-Simulation of Hydrogels

• Solvents: Water, Ethanol, aqueous NaCl solution• Temperatures: 260 K - 340 K• Force fields from literature

PVA PNiPAM PAA


39

MD-Simulation: Collapse of Hydrogel

primitive PVA-network


Legend:Default Water model of force field

- Temperature dependence not observed+ Temperature dependence observable++ Temperature dependence reasonably predicted

Force Field Study

PNiPAM Water: SPCE Water: TIP4PGromos96 UA - -Gromacs53a6 UA - +OPLS AA ++ +


41

MD-Simulation PNiPAM-Chains

T < TΘ

T > TΘ


Survey of Coordinated Research ProgramsDFG SPP 1155:Molekulare Modellierung und Simulation in der VerfahrenstechnikProcessNet Arbeitsausschuss MMS:Molekulare Modellierung und Simulation für das Prozess- und ProduktdesignDFG SFB 716:Dynamische Simulation von Systemen mit großen TeilchenzahlenDFG TFB 66:Molekulare Modellierung und Simulation zur Vorhersage von Stoffdaten für industrielle Anwendungen BMBF IMEMO: Innovative HPC-Methoden und Einsatz für hochskalierbare Molekulare Simulation

SFB 716


SummaryMolecular Modelling and Simulation in Process Engineering

used in industry high potential is recognizedfuture key technologytruly interdisciplinary field

Co-operation between EngineeringNatural ScienceComputer ScienceMathematics


Thanks to co-workers…

Jürgen StollThorsten Schnabel Gimmy FernandezBernhard EcklIsaiah HuangMartin HorschThorsten MerkerGabriela GuevaraJonathan WalterStephan DeubleinCemal Engin

…and colleagues from industryJohannes Vorholz (Evonik)Robert Franke (Evonik)Bernd Eck (BASF)Manfred Heilig (BASF)

ASIM Workshop on Fundamentals of Modeling and Simulation...

Documents

Transcript of ASIM Workshop on Fundamentals of Modeling and Simulation...