![]() The three xyz-files each contain 16 benzene molecules whose latticeĬonstants have been equilibrated at 250, 300 and 350 K. Viscosity.py and the xyz files for 250 K, 300 K, 350 K. We therefore recommend you to read off the value manually (and not simply However,Įxtracting the converged viscosity value can be difficult to do automatically. (ACF) integral can be scripted with our PLAMS scripting framework. (the above example used the experimental density, but that may not be suitable for the used force field)īetter and more reliable viscosity values can most likely be obtained using the APPLE&P or GFNFF force fields.īoth the MD simulation and the calculation of the autocorrelation function In general, make sure to equilibrate the density of the liquid first using an NPT simulation, before running an NVT simulation as above. ![]() It is therefore necessary set the Sample frequency to a small number, ideally even smaller than the 5 fs we used here. The biggest contribution to the integral comes at short correlation times. You may want to increase the maximum correlation time (above given as 3000 frames or 15 ps) to a larger value, and increase the simulation time. The integral must converge in the long time-limit. Use a larger supercell with more benzene molecules for a better value. ![]() The viscosity is often sensitive to the supercell size. The simulation needs to be well equilibrated The experimental room-temperature viscosity of benzene is about 0.5 mPa s. Pardon my laziness for asking instead of digging through the documentation, but it is possible to tweak the mixture model (change of reference, for instance)? Pointers to applicable starting points would be appreciated.\ (Why don't I just use this model, you may ask? TREND have problems with viscosities under other pressure/temperature states this is one where TREND worked, but CoolProp didn't) I don't even know how you would setup the apparatus to get some measures under these conditions. Red and green are pure viscosities as reported by CoolProp. mixture model of corresponding states (originally by al-Siyabi for CO2 (2013), from Heriott-Watts), but with Octane as reference component (!), because it has the most mass in this mixture, coded up with a version of TREND before 3.0 (!), I am able to produce this: I think we are narrowing it down to the selection of mixture model and/or calibration of it, and not some artifact numerical error (non-convergence, for instance). I would try to do some better attempt at debugging, but the numbers after CoolProp.cpp in the error message certainly doesn't seem to be valid line numbers in the source code (maybe after being run through the preprocessor, though). If I am to speculate wildly, viscosity is probably calculated using some kind of corresponding state model, and if CO2 (or some other light component) is used as a reference, maybe the mass of the Octane component is relatively large, such that the corresponding state is in the solid phase(!). # CoolProp/CoolProp.pyx in CoolProp.CoolProp._Props_err2 (CoolProp/CoolProp.cpp:39762)() # CoolProp/CoolProp.pyx in (CoolProp/CoolProp.cpp:41504)() # CoolProp/CoolProp.pyx in (CoolProp/CoolProp.cpp:41716)()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |