Page 121 - Intelligent Digital Oil And Gas Fields

88 Intelligent Digital Oil and Gas Fields

• Fig 3.3B shows a well in unsteady state condition with constant water
rate. The data is replaced by an extrapolation/interpolation approach.
• Fig 3.3C shows a well response from changing the choke size with con-
stant water cut. The gas-rate data can be replaced by using a predictor
such as a trained ANN, FzL, or nodal analysis, before and after changing
the choke size.
• Fig 3.3D shows a well during a test with different pump frequencies,
from low to high frequencies, increasing oil rate, and decreasing flowing
bottom-hole pressure (fBHP). In this case, the fBHP is missing, so a 1D
analytical model calibrated with reservoir properties can replace the
fBHP data.
• Fig 3.3E shows a well with declining oil rate. fBHP is approximately
constant because water injection maintains the reservoir pressure. Water
breakthrough increases water cut from 20% to 40%; so in this situation, a
3D numerical model could be the best tool to replace data. An ANN can
be trained with a 3D numerical model to predict water and oil rate data
instantaneously, as observed in Fig. 3.7F.
• Fig 3.3G shows an unconventional well (low permeability) fractured
with slick water, as observed water produces first at high volume, and
gas increases and then decreases when a boundary condition is hit.
Gas rate and water could be replaced between flow regimes with rate
transient analysis (RTA) type curves.
• Fig 3.3H shows a scenario where a well produces under critical
condition or water loading up; in this case replacing data is a challenge
and might not be practical or appropriate.

3.2.5 Data Reconciliation
Production reconciliation is a data process and statistical method to calculate
a final production value when two or more different sources and measure-
ments are available. In DOF systems, it is common that two or three meters
can easily mismatch and that not all devices or methods measure production
correctly at all times. Quite often, the dispatcher and receiver have a signif-
icant discrepancy in fluid readings. Owing to these conditions, the DOF sys-
tems require reconciliation methods to correct input data and generate
unique readable, cleaned-up, and validated output information. Reconcil-
iation is used to match fluids (e.g., gas, water, and oil) but rarely pressure.
Like any other statistical method, the reconciliation process can match only
the data within the preset limit, tolerance, and uncertainty values. The limits