Interpretation of radioactive marker measurements in the Northern
Adriatic gas fields
M. Ferronato, G. Gambolati, P. Teatini
Dept. Mathematical Methods and Models for Scientific
Applications, University of Padova, Padova, Italy
D. Bau'
Dept. Geological and Mining Engineering and Sciences, Michigan Technological
University, Houghton, Michigan, USA
ABSTRACT
The radioactive marker technique (RMT) appears to be a promising
tool to evaluate the in-situ uniaxial vertical compressibility
cM of deep producing gas/oil reservoirs. However,
when the field consists of multipay layers with the porous-medium
heterogeneity scale smaller than the marker spacing (10.5 m),
great care must be exercised in the interpretation of an RMT survey.
If the monitored depth interval incorporates thin clayey layers
or the measurement is made in an active pumping well,
cM can be grossly underestimated. Moreover,
rock may expand, with the expansion correctly recorded by RMT.
In the present paper, a set of RMT measurements made by Eni-E&P
over the past decade in three deep boreholes of the Northern
Adriatic basin are simulated with the aid of a 3D poroelastic
model solved by finite elements (FEs). Use is made of the much
detailed lithostratigraphies of the test holes and the
cM constitutive equation derived from a previous
statistical analysis of the marker data. The modeling results show
that the measurements also can be reproduced satisfactorily from a
quantitative viewpoint, and they indicate that an efficient marker
installation requires that the monitored depth interval be made
mostly from an entirely depleted sandy unit, with the markers
placed possibly far from a producing well approximately 10.5 m
apart (i.e., the distance between the gamma-ray detectors which
monitor the marker position). The cM constitutive
law used in the numerical analysis is realistically accurate from
2500 to 3000 m depth, while it appears to be underestimated by a
factor of 2 between 800 and 1500 m. The measured expansions allow
for the assessment of the reservoir cM under
unloading conditions. This turns out to be 2.5 times smaller than
cM in virgin loading conditions.
