Improved correlations calculate diesel cold-flow properties from compositional properties of the fuel's blending components. The values calculated by multiple regression equations are within 1% of the corresponding experimental values.

H.U. Khan, a scientist at the Indian Institute of Petroleum, Dehradun, derived the correlations using diesel-blending components from Bombay High crude oil. Khan disclosed the results of his tests in an unpublished report.

The deviations achieved are well within the repeatability limits of the ASTM/IP tests for the properties studied, indicating the new correlations are better than those reported earlier, according to Khan.

### EXISTING CORRELATIONS

Several graphic and mathematic correlations have been suggested to calculate diesel-blend properties from known component values.13 These studies, except that of Krishna, says Khan, are either specific in their use or valid for only a limited number of blending components (3 or 4).4

An analysis of Krishna's results, however, indicates that, for more than 50% of the data, the variation between experimental and calculated values is large, says Khan.

Furthermore, no procedures have been suggested to calculate the compositional properties of the blends from the known component values. This, says Khan, reduces the applicability of the proposed correlations.

In his work at the Indian Institute of Petroleum, Khan considered linear correlations, both logarithmic and multiple-regression type, with an increased number of interaction terms. The multiple regression equations proved superior.

An analysis of Krishna's data for Bombay High yielded an equation for calculating the compositional properties of the blend in terms of those of the blending components:

[see formula]

(See Nomenclature in Table I below.)

### NEW CORRELATIONS

The cold flow properties studied were cloud point, pour point and cold filter plugging point (CFPP), all in C. The compositional properties used in the correlations include: ASTM midboiling temperature, or T50; wax content; paraffin content; and average carbon chain length (Nomenclature).

The compositional variables have both primary and interaction effects on the cold-flow properties. The total number of interaction parameters is 18.

For each cold flow property with a varying number of interaction terms (up to a maximum of 15), a total of 34 polynomials were derived. The equations with 15 variables were well within acceptable limits, producing an average R2 greater than 0.99 and Yest. less than 1.52.

The 15-variable equations are:

[see formula]

For diesel oil blends of known volume composition, the values of XI, X2, X3, and X4 are calculated using Equation 1. The values then are substituted into Equations 2, 3, and 4, to calculate cloud point, pour point, and CFPP.

### VALIDITY

Khan checked the validity of Equations 2-5 by using experimental values of X1-X4, or values calculated using Equation 1, to calculate cloud point, pour point, and CFPP. Fig. 1 shows how these calculated cold-flow properties compare to experimentally derived properties for 28 blends.

Table 1 shows that the great majority of the calculated data deviate from the experimental values by less than 1%.

### REFERENCES

- Khan, H.U., Mungali, M.M., Agrawal, K.M., and Joshi, G.C., "Graphical method simplifies diesel cloud point determination," OGJ, Sept. 24, 1990, p. 98.
- Khan, H.U., Mungali, M.M., Agrawal, K.M., and Joshi, G.C., "Graphical correlation calculates diesel properties from known values," OGJ, Sept. 23, 1991, p.47.
- Tsang, C.Y., Ker, V.S.F., Miranda, R.D., and Wesch, C.J., "Equation predicts diesel cloud points," OGJ, Mar. 28, 1988, p. 33.
- Krishna, R., Bhattachajee, S., Joshi, G.C., Singh, H., Purohit, R.C., Dilawar, S.V.K., and Singh, K.K., "Correlation of low temperature flow properties of diesel fuel with composition," Erdol, Kohle, Erdgas, Petrochem., 42(2), 1989, pp. 72-75, and Hydrocarbon Technology, February 1989, pp. 37-49.

*Copyright 1994 Oil & Gas Journal. All Rights Reserved.*