Page 192 - Machine Learning for Subsurface Characterization

166 Machine learning for subsurface characterization

can achieve miscibility with the injected hydrocarbon, while the hydrocarbon
residing in larger pores remains immiscible. The lack of miscibility results in
interfacial tension and multiphase flow that reduces the displacement
efficiency achieved by the injected light hydrocarbon.
In shales, the free-oil volume cannot be considered as a direct indicator of
microscopic displacement efficiency. Instead the fraction of free-oil volume
that can achieve miscibility is a better indicator of the displacement
efficiency in shales. The presence of movable water in the pores will
adversely affect the displacement efficiency. However, bound water has
complex effect on light-hydrocarbon injection. Several studies have come up
with contradictory results [30]. Here, we assume that bound fluid has a
negative effect on the displacement efficiency. Furthermore, kerogen content
increases pore complexity, and pores are preferentially oil wettability,
thereby reducing the displacement efficiency. Another adverse factor is the
presence of bitumen. Dominant pore throat size is 5 nm in the upper and
lower shales and 25 nm in the middle shale [14]. Bitumen have diameters
ranging from 5 to 200 nm [31]. Bitumen will tend to clog pores and inhibit
flow in the shales. Such complex interactions and dependencies mandate the
construction of an index to better quantify the complex process of oil
recovery due to light hydrocarbon injection.

4.2 Calculation of the MD-index

The calculation process for the MD-index is shown in Fig. 6.3. MD-index
incorporates volume fraction of miscible free oil, bound fluid, movable
water, and kerogen. MD-index is formulated as a simple ratio of positively
affecting parameters to the adversely influencing parameters expressed as
V o
I MD ¼ (6.5)
ð V w + V b Þ V k
where V o , V b , and V w is the pore volume fractions of miscible free oil, bound
fluid, and movable water with respect to the bulk volume, V k is the kerogen
volume fraction with respect to bulk volume, and I MD is the MD-index. The
technical challenge is to accurately estimate the aforementioned properties.

FIG. 6.3 Procedure for the MD-index calculation.