PROPERTIES OF RESERVOIR GASES - Petroleum Reservoir Fluid Property Correlations - page 10

Table 2-2 shows that the Hall-Yarborough representation of the
Standing-Katz z-factor graph gives slightly better results than the DAK
equations. Figure 2-1 also shows that the Hall-Yarborough equations
give slightly better results with the data set divided into 11 approximately
equal-sized subsets according to gas specific gravity.
The comparisons in figure 2-1 used the Piper et al. equations
to calculate the pseudocritical properties. Comparisons of the
Hall-Yarborough equations with the DAK equations using other
pseudocritical correlations and with the data sorted by temperatures
and pressures gave similar results. Elsharkawy et al. reported a similar
conclusion. 14
An evaluation of the results of these procedures using routine
statistical techniques showed that there is no difference (in a statistical
sense) between the results from Hall-Yarborough and DAK. Both sets
of equations require very difficult trial-and-error solutions; occasionally
the Hall-Yarborough solutions did not converge using either the
Hall-Yarborough procedure or the fairly robust Dake procedure. 15 Thus,
the DAK equations will be used in further evaluation of the correlations
for determining pseudocritical properties.
The Standing modification of the Beggs-Brill equations (SBB) gave
results nearly as good as the Hall-Yarborough and DAK methods. This
SBB procedure is recommended for use in situations in which trial-
and-error solutions are not feasible. The Papay equation gave very poor
results, as was reported by Takacs.
Table 2-2 also shows that the Piper et al. correlation equations and the
Sutton (2007) correlation (his equations 10 and 11) for pseudocritical
properties give the more nearly accurate predictions of z-factors. This
table gives results for the entire data set. An examination of subsets of
these data will show how well the correlations predict z-factors across
the ranges of temperature, pressure, and gas specific gravity that might
be encountered in application.
Table 2-3 gives a comparison of the methods of estimating the
pseudocritical properties for four subsets of the data containing
appreciable amounts of nonhydrocarbon components. The Sutton
(2007), Piper et al., and Standing correlations are approximately as
accurate for gases with nonhydrocarbon components as for sweet gases.
The poor results for the Sutton (1985) correlation are not surprising; the
gas samples in his data set had no hydrogen sulfide and limited amounts
of carbon dioxide and nitrogen. Elsharkawy et al. did not mention using
the Wichert-Aziz adjustments for hydrogen sulfide and carbon dioxide;
however, these adjustments were tried and did not improve the results.