Pure Property Regression

The regression add-on for pure component properties allows for extended parameter fitting, calculation and graphical representation options for various correlation equations for the different pure component properties that allow customization of simulation software for a better description of your specific mixtures.

Parameters can be stored in and retrieved from a parameter database, they can be plotted, and they can be used for calculations. DDBSP Pure Property Regression normally uses the pure component properties data bank which is a part of the Dortmund Data Bank. It can also be used to fit data from other data sources since tables can be pasted from the clipboard or loaded from files. The main dialog of the program is shown in the following figure:



After selecting the component as well as the property and equation of interest, a regression dialog displays several additional options like restriction to certain temperature or pressure ranges, type of objective function and the inclusion of additional boundary conditions like the Riedel criterion or the correct location of the minimum in slope. Choosing either nonlinear or multilinear fit, the following results are shown:



In most cases, several regressions are performed with either varying number of parameters or, as in this case, with varying exponents in the last term of the DIPPR 101 equation. Selecting plot leads to a dialog with various graphical representations:



When moving the mouse curser over the diagram, a connecting line is drawn to the nearest data point and the reference and exact numerical values are shown. Chosing the context menu (right mouse button) allows to exclude this data point or all data points from the resp. reference from the next regression. Excluded data points are shown in a different color for convenience.

Regression results can be exported or stored in the ParamDB data bank. I addition, information about time of the fit and all included and excluded data can be stored in an archive, so that these selections can be reviewed at any later time. DDBSP Pure Property Regression is specifically targeted at rapid and interactive regression of reliable parameters for different pure component properties.