Fitting order for properties

AVEVA™ Assay uses an ordering when calculating the value for properties:

1. a property is part of the statistical model, the result is calculated from the modelled representation

2. if a property is not part of the statistical model, and sufficient measured values are available, they will be fitted using a curve fitting method. Usually at least 2 values, or one value and a constraint are required to define the curve

3. if a property is not part of the statistical model, and no measurements are available, any correlations available will be used to calculate a profile. See Specifying Constraints in Configure your properties.

4. if the property is not part of the statistical model, no measurements are available, and no custom correlations are available, no values will be calculated

Therefore if sulphur values are entered for an assay as sulphur is a modelled property then profile fitting is performed using the entered values and the statistical model. If no sulphur values were entered a sulphur profile would be generated using the statistical model.

If flash point values were entered for several streams (flash point is a correlated property) then the profile would be fitted using the values and any applied constraints. If no flash point values were entered the flash point would be calculated using the correlation.

If arsenic values were entered for several streams (arsenic is not modelled and not correlated) then the profile would be fitted using the entered values. If no arsenic values were entered no profile would be generated.

Warning: If you enter one or more values for a correlated property these values will be used to calculate the profile rather than the correlation.
As a result if you only enter a few data points, or these points are not accurate the calculated profile may not be correct, or in the case of one data point appear as a straight-line. In this case the entered values should be deactivated in the Measurement Grid and Assay allowed to use the standard correlation to predict the profile.
For example, if you enter a single flash point value this is used, along with the flash point constraints to draw a profile, but the resultant profile is a flat line because there is only a single data point. To generate a correct profile for flash point either enter more data for other streams or deactivate the flash point measurement and allow the correlation to be used to calculate the flash point profile.