2 replies to this topic

[waheed2020]

Dear All,

 

I did the characterization of pyrolysis oil.  2D GC analysis ASTM D2887, TBP, Density, viscosity at 70 and 70. Then simulate the data to perform fractionation using batch aspen plus simulation. The data matched with the experimental batch setup. Mass balance, purity of fractions, and density. However, two values of viscosity are out of the range predicted by the model. 

 

 

My question is how can I calibrate the viscosity in ASPEN plus?

 

Thanks in advance for help

 

 

[ProcessAtWork]

This is actually the expected failure mode rather than a sign of an error: density tracks boiling point and specific gravity very faithfully, so the standard pseudocomponent correlations match it well, whereas viscosity is exponentially sensitive to molecular structure and intermolecular forces (H-bonding, aromatics, oxygenates) that the petroleum-based correlations in Aspen don't capture for pyrolysis oil. So this is a calibration problem, not a model-physics issue. The main levers are: make sure your measured kinematic viscosities were actually entered into the Petro Characterization assay (if only D2887/TBP and density went in, Aspen estimated viscosity from correlations and never saw your lab data); check the liquid viscosity mixture method under Transport Properties; and, for the two off fractions specifically, regress the Andrade parameters (MULAND) against your fraction-level viscosity–temperature data. One thing to watch is extrapolation — Aspen fits viscosity vs. temperature with the Walther equation, and with only two measured points it extrapolates poorly outside that bracket, so if the off fractions are being evaluated outside your measurement range, that alone can explain the deviation.

 

That last point is really the crux: the fit is only as good as the number of viscosity-temperature points you feed it, so the cleanest fix for those two fractions is more data across temperature rather than a different correlation. Using an online viscometer makes this practical: you can capture a continuous viscosity-temp curve on each fraction directly in the batch setup instead of relying on two discrete lab points, which gives Aspen a properly bracketed dataset to regress against and also rules out measurement artifacts and any water/emulsion effects, since you're reading the actual process fluid at process conditions.

[latexman]

waheed2020 was “Last Active Jan 22 2024 04:39 AM”