60 |
Oilfield Technology
August
2015
system at a later time, if required. This capability makes it possible to
build up a collection of calibration parameters tailored for different
situations, and the FSE can then choose the coefficient that gives the
best performance in the current situations.
Challenges inMPT
When the MPT telemetry systemwas introduced it provided data
rates faster than 6 bits per second. Over the course of time, as the
market demand for high-speed telemetry increased, several additional
challenges surfaced.
Ì
Ì
More demanding mud properties and deeper wells required manual
assistance to detect the TS, putting a higher workload on the FSE.
Several factors contribute to this and must be addressed:
h
h
Some frequency contents of the TS can be strongly attenuated
by the mud system. This situation mainly happens when using
mud systems with a high PV or in very deep wells.
h
h
The training sequence can be distorted by downhole or
surface noise that cannot be filtered out by the system. Pump
noise, in particular, can be a problem under certain conditions.
h
h
The system is designed for the use of two pressure transducers
(PTs). On many rigs there are only two, but some would
benefit if the processing were performed on more available
PTs.
h
h
Manual assistance when the automatic TS detection fails is
cumbersome, and requires more training and attention from
the FSE.
h
h
The search requires a manual trigger. Automation would
remove this task from the FSE, giving time for more valuable
tasks.
Ì
Ì
The manual fallback options to calibrate the processing algorithms
require a high level of training to use them effectively. Only a small
percentage of all FSEs can use them. Most FSEs do not attempt to
use them, and instead choose to use a slower data rate.
Ì
Ì
Performing a new calibration by means of TS or fallback options
can only be done on the live system, and the new settings
are immediately applied. Field operations have shown that a
pre-testing of coefficients prior to their operation is needed.
Ì
Ì
Managing the acquired calibration parameters is difficult with their
rising number. It is not unusual to have several saved calibration
parameters that are used alternately, e.g., for different flow rates
and pump combinations, to reach optimum decoding. Problems
frequently arise from finding the correct coefficient file for the
current situation. A parameter management system is needed.
Ì
Ì
As modern drilling rigs appear, new mud pumps types and
dual derrick setups show that the surface mud flow line-up is
getting increasingly complex and requires rigging up several PTs
for decoding. On some installations up to 6 PTs are installed in
different locations to switch between different lines. The correct
combination of PTs has a significant influence on the decoding
performance, and finding the best combination on the live system
requires time. An automated, offline evaluation of the possible
PT combinations is needed.
Newsystemimprovements
After gaining much experience in the field, data sets were analysed
and user experiences were collected to improve the current system.
As a direct answer to the limitations, several major and minor
improvements were introduced to reach higher data rates and
improved robustness.
Ì
Ì
The detection of the TS was drastically improved. Several
improvements contributed to this achievement:
h
h
For deep wells and high-viscosity mud, the automatic search
of the TS was less reliable. The TS was expanded to lower
frequencies to reach higher robustness for those wells.
h
h
An improved noise cancellation technique was applied before
the TS was detected. This improved reliability.
h
h
The TS detection was expanded to all available pressure
channels, not just the ones used for decoding.
h
h
A manual selection possibility was added to the new system if
the automation failed.
h
h
The TS detection was automated, releasing the FSE from
performing the manual trigger.
h
h
The efficiency of the TS was improved through educated
changes, leading to shorter TS with improved decoding
quality.
Ì
Ì
The new system improved the reliability of the TS detection
tremendously for previously challenging conditions, eliminating
most situations where manual interaction was needed. Cases
requiring manual user selections were simplified in an easy-to-use
selection window where the user had to look for a characteristic
pattern.
Ì
Ì
New coefficient sets were tested on real data to generate
a decoding quality prediction. Changing conditions during
the transmission of the TS led to different qualities of the
coefficients. The user could select to apply coefficients based
on their performance. This avoided applying non-optimal sets
of coefficients directly to the real time system. In addition the
FSE would gain additional confidence in picking a good set of
coefficients.
Ì
Ì
A database of coefficient sets was created to assist the FSE
through automated saving of the sets with corresponding
additional information such as rig status, flow rates, active pumps
or decoding parameters. This further simplified the coefficient
selection to best match the current conditions.
Figure 4.
Spectrogramof pressure signal whileBHA is at shallow
hole depth. Reddepicts strong signal content, blue showsweak signal
content.
Figure 5.
Pressure signal before andduring high stick/slip in time domain.
