SPWLA Twenty-Second Annual Logging Symposium, JUNE 23-26, 1981        PAPER J

PAPER J

 

COMPUTER SAND COUNTING AND CURVE SQUARING

 

B. H. Bailey and M. de Crespo

Lagoven, Tia Juana, Venezuela

 

ABSTRACT

 

For the last ten years Lagoven has been using computer generated sand counts for reservoir description projects. It became obvious in 1968 that the large 1100 well reservoirs developed over a long period of years had inconsistent sand counts. The definition of net oil sand changed with time, economics, personnel, so that comparable wells drilled 20 years apart could have NOS values that are different by as much as 50%. It was decided that it would be impossible to recount over 4,000 well logs by hand so a computer program was developed (SANDAN) to systematize the net sand determination. The count program inputs two digitized curves such as the SP and short normal and outputs the zonation of the section as well as a determination of which zones are sand and which are shale. Base line values are also established such as the resistivity of the shale (Rsh), sand/shale resistivity cutoff, SP shale and SP sand. If the SP curve is not useable as determined by cross correlation, the GR may be substituted. The resistivity side does not have to be the short normal but this is the usual curve in as much as it is common to all wells old and new. Experience has shown the sand count based on the short normal very closely matches the hand count if all the sand is counted. Later, cut-offs can be made so as to convert net sand into net oil sand.

 

After the well is zoned the program reconstructs, i.e. “squares”, the short normal by correcting for thin bed and bore-hole effects. Each zone then becomes by definition a homogeneous vertical unit, that is, all petrophysical parameters are now taken to apply through the entire zone. If the zonation is fine enough this is usually true. How fine should the zonation be? We want the Rt, porosity, shale content, etc., to be constant for the zone extent. Experience has shown that SANDAN normally will result in 1.5 to 2 times as many zones as will be done by hand. Significant variations within the zones of the other log values are seldom found.