Accelerating Monte Carlo molecular simulations by reweighting and reconstructing Markov chains: Extrapolation of canonical ensemble averages and second derivatives to different temperature and density conditions

Handle URI:
http://hdl.handle.net/10754/563666
Title:
Accelerating Monte Carlo molecular simulations by reweighting and reconstructing Markov chains: Extrapolation of canonical ensemble averages and second derivatives to different temperature and density conditions
Authors:
Kadoura, Ahmad Salim ( 0000-0001-9317-682X ) ; Sun, Shuyu ( 0000-0002-3078-864X ) ; Salama, Amgad ( 0000-0002-4463-1010 )
Abstract:
Accurate determination of thermodynamic properties of petroleum reservoir fluids is of great interest to many applications, especially in petroleum engineering and chemical engineering. Molecular simulation has many appealing features, especially its requirement of fewer tuned parameters but yet better predicting capability; however it is well known that molecular simulation is very CPU expensive, as compared to equation of state approaches. We have recently introduced an efficient thermodynamically consistent technique to regenerate rapidly Monte Carlo Markov Chains (MCMCs) at different thermodynamic conditions from the existing data points that have been pre-computed with expensive classical simulation. This technique can speed up the simulation more than a million times, making the regenerated molecular simulation almost as fast as equation of state approaches. In this paper, this technique is first briefly reviewed and then numerically investigated in its capability of predicting ensemble averages of primary quantities at different neighboring thermodynamic conditions to the original simulated MCMCs. Moreover, this extrapolation technique is extended to predict second derivative properties (e.g. heat capacity and fluid compressibility). The method works by reweighting and reconstructing generated MCMCs in canonical ensemble for Lennard-Jones particles. In this paper, system's potential energy, pressure, isochoric heat capacity and isothermal compressibility along isochors, isotherms and paths of changing temperature and density from the original simulated points were extrapolated. Finally, an optimized set of Lennard-Jones parameters (ε, σ) for single site models were proposed for methane, nitrogen and carbon monoxide. © 2014 Elsevier Inc.
KAUST Department:
Computational Transport Phenomena Lab; Physical Sciences and Engineering (PSE) Division; Environmental Science and Engineering Program; Chemical and Biological Engineering Program; Earth Science and Engineering Program
Publisher:
Elsevier BV
Journal:
Journal of Computational Physics
Issue Date:
Aug-2014
DOI:
10.1016/j.jcp.2014.03.038
Type:
Article
ISSN:
00219991
Sponsors:
The work presented in this paper has been supported in part by the project entitled "Simulation of Subsurface Geochemical Transport and Carbon Sequestration" (award number 7000000058), funded by the GRP-AEA Program at KAUST.
Appears in Collections:
Articles; Environmental Science and Engineering Program; Physical Sciences and Engineering (PSE) Division; Chemical and Biological Engineering Program; Earth Science and Engineering Program; Computational Transport Phenomena Lab

Full metadata record

DC FieldValue Language
dc.contributor.authorKadoura, Ahmad Salimen
dc.contributor.authorSun, Shuyuen
dc.contributor.authorSalama, Amgaden
dc.date.accessioned2015-08-03T12:05:32Zen
dc.date.available2015-08-03T12:05:32Zen
dc.date.issued2014-08en
dc.identifier.issn00219991en
dc.identifier.doi10.1016/j.jcp.2014.03.038en
dc.identifier.urihttp://hdl.handle.net/10754/563666en
dc.description.abstractAccurate determination of thermodynamic properties of petroleum reservoir fluids is of great interest to many applications, especially in petroleum engineering and chemical engineering. Molecular simulation has many appealing features, especially its requirement of fewer tuned parameters but yet better predicting capability; however it is well known that molecular simulation is very CPU expensive, as compared to equation of state approaches. We have recently introduced an efficient thermodynamically consistent technique to regenerate rapidly Monte Carlo Markov Chains (MCMCs) at different thermodynamic conditions from the existing data points that have been pre-computed with expensive classical simulation. This technique can speed up the simulation more than a million times, making the regenerated molecular simulation almost as fast as equation of state approaches. In this paper, this technique is first briefly reviewed and then numerically investigated in its capability of predicting ensemble averages of primary quantities at different neighboring thermodynamic conditions to the original simulated MCMCs. Moreover, this extrapolation technique is extended to predict second derivative properties (e.g. heat capacity and fluid compressibility). The method works by reweighting and reconstructing generated MCMCs in canonical ensemble for Lennard-Jones particles. In this paper, system's potential energy, pressure, isochoric heat capacity and isothermal compressibility along isochors, isotherms and paths of changing temperature and density from the original simulated points were extrapolated. Finally, an optimized set of Lennard-Jones parameters (ε, σ) for single site models were proposed for methane, nitrogen and carbon monoxide. © 2014 Elsevier Inc.en
dc.description.sponsorshipThe work presented in this paper has been supported in part by the project entitled "Simulation of Subsurface Geochemical Transport and Carbon Sequestration" (award number 7000000058), funded by the GRP-AEA Program at KAUST.en
dc.publisherElsevier BVen
dc.subjectCanonical ensembleen
dc.subjectLennard-Jones modelen
dc.subjectMarkov chain reweighting and reconstructionen
dc.subjectMolecular simulationen
dc.subjectMonte Carloen
dc.titleAccelerating Monte Carlo molecular simulations by reweighting and reconstructing Markov chains: Extrapolation of canonical ensemble averages and second derivatives to different temperature and density conditionsen
dc.typeArticleen
dc.contributor.departmentComputational Transport Phenomena Laben
dc.contributor.departmentPhysical Sciences and Engineering (PSE) Divisionen
dc.contributor.departmentEnvironmental Science and Engineering Programen
dc.contributor.departmentChemical and Biological Engineering Programen
dc.contributor.departmentEarth Science and Engineering Programen
dc.identifier.journalJournal of Computational Physicsen
kaust.authorKadoura, Ahmad Salimen
kaust.authorSun, Shuyuen
kaust.authorSalama, Amgaden
All Items in KAUST are protected by copyright, with all rights reserved, unless otherwise indicated.