Page 235 - Intelligent Digital Oil And Gas Fields

Workflow Automation and Intelligent Control 185

the engineers can update the nodal analysis model and compare the result
with the previous available test analysis and identify significant changes in
IPR or VLP. On the basis of our work on this project, we know that the
automated workflow can tune the model parameters until the acceptable
error is below 10% between the observed gas rate and f BHP , versus simulated
gas rate and f BHP . This workflow displays the multiphase flow equation
through time in an iterative GIS map, which shows several wells with dif-
ferent correlations. Ultimately, the physical model is used to validate the well
test. If the well test data matches the well model with an error <10%, test is
considered validated and accepted; otherwise, the test is rejected and must be
repeated.
However, this automated workflow requires the integration of machine
learning to memorize previous tuning steps and to be consistent throughout
the production history. Normally, the tuning process uses basic equations,
such as Vogel, Darcy, or flow parameters such as c and n factors to calibrate
the IPR curve. The PEs should have the reservoir pressure, skin, and matrix
permeability data; however, if this data is not available, the engineers can
change these properties until there is a minimum error, but sometime these
changes are meaningless. To avoid meaningless changes, machine learning
canbeusedtomemorizethechangesinreservoirpressureduringthereservoir
depletion or provide this value from material balance and numerical models.
The diagnostics provided by the automated workflows need to be
reinforced with expert rules analysis, fuzzy logic, and management-by-
exception rules. These techniques improve diagnostics of the well trouble-
shooting and provide accurate recommendations for further action, for
example:
• If the stability check KPI is >0.0, the expert rule recommendation could
be to reduce the choke size and stop the test.
• If data is frozen, then generate an alarm.
• If WC or GOR increases with a multi-rate test, then suggest the best
choke size.
• If gas rate and f BHP do not match, then tune the multiphase flow
correlation.

5.6 DIAGNOSTICS AND PROACTIVE WELL
OPTIMIZATION WITH A WELL ANALYSIS MODEL

Diagnostics and well optimization are routine activities performed
daily by PEs. This section describes typical diagnostics and procedures