Auburn UniversityRobert Kline
1/19/12
Why? How?
• Data Regression• Analogous Systems• Estimation
Pure component data exists for most commercially interesting chemicals• Can be estimated for new chemicals
Mixture data is available for a small fraction of all binary pairs• Interaction parameters (activity and
fugacity coefficients) are critical model chemical processes
Sources of data• Journals• Compilations (DECHEMA, Horsley)• Internal Reports• Measurements
Data quality• Not all data is created equally• GC Technology (1970s +) greatly increased
liquid composition accuracy
When should you have it measured?• If it is a critical split between important
chemicals Products with close boiling components Key splits in columns If the presence of an azeotrope will render a
separation infeasible
How to regress?• Use ASPEN• Set mode to Data Regression under Setup• Create a new Data Set under Properties• Create a new Regression Set under
Properties• Run the Regression• Set mode back to Flowsheet Simulation
under Setup
Sometimes an unknown system is like a known system (butyl acetate + water vs isobutyl acetate + water)
Two options• Use the same parameters (if the boiling
points are similar)• Use the same infinite dilution activity
coefficients and regress new parameters
UNIFAQ• Group contribution method
• Breaks down the molecule into chunks called groups• Calculates activity coefficients based on group
concentrations instead of molecule concentrations • Estimates data and regresses your parameters• Can be reasonably accurate, but you should always
review the results to make sure it makes sense
H3C CH2 CH2
OHH3C O C CH3
OGroup # in A # in B 50/50
Mix75/20Mix
Alkane 2 3 2.5 2.25
Ester 1 0 0.5 0.75
Alcohol
0 1 0.5 0.25