Geoscience Reference
In-Depth Information
Evaluation Criteria
The evaluations were based on
average relative error
and
average absolute
relative error
as defined below.
Average relative error, ARE, %:
(1.1)
Average absolute relative error, AARE, %:
(1.2)
ARE is simply the arithmetic average of the relative errors. It is a
measure of the bias of the correlation; a value of zero indicates a random
deviation of the measured values around the correlation. If the deviations
in the laboratory-measured values of a property are truly random, a
correlation with a very small value of ARE will result in predicted values
as accurate, or possibly more accurate, than laboratory measurements.
AARE is the arithmetic average of the absolute values of the relative
errors; an indication of both the precision of the correlation and the
accuracy of the data. A small value of AARE denotes a good correlation
based on good data. A large value of AARE could mean a poor quality
correlation (inadequate functional form). However, the situation of
several different correlation equations all with large values of AARE
most likely indicates poor quality data.
ARE and AARE are just more robust versions of the more familiar
statistical measures, mean and standard deviation. Their use here
is justified because they require fewer assumptions regarding the
distribution of the measurement errors.
Evaluation Procedures
The several correlation equations proposed in the petroleum
literature for each fluid property were evaluated using appropriate sets
of data. These sets of data are fairly large and are based on reservoir fluid
Search WWH ::
Custom Search