Geology Reference
In-Depth Information
10.1.5 Surface signal processing.
The depth over which MWD transmissions can operate successfully
depends not only on signal strength created at the source, but importantly, on the
“signal to noise ratio” (S/N) found at the surface. To emphasize this point, we
suggest that a small 0.1 psi signal in itself might not be entirely detrimental if
noise did not exist. But it does and, very often, overwhelms the upcoming
signal. Different types of noise are found at the surface, e.g., mud flow noise of
a random nature, noise associated with drillstring vibrations and rig operations,
and so on, many of which can be removed using conventional frequency
filtering methods.
However, a major source of problems is propagating noise traveling in a
direction opposite to the upcoming signal, e.g., MWD signals reflecting from
mudpump pistons, shape-distorted signals reflected from the desurger and rotary
hose, and very large noise amplitudes created by moving duplex and triplex
pump pistons themselves. In principle, these are filtered by multiple transducer
signal processing methods, but several service-company schemes used to this
author's knowledge are derived under dubious assumptions. For instance, some
unrealistically assume sinusoidal time variations, while others casually invoke
“common sense” subtraction methods. None apply the degree of rigor found in
seismic processing, which is based on exacting geophysical models.
The multiple transducer methods in Chapter 4, however, are based on
formal wave equation manipulations and results. For instance, in Method 4-4,
the one-dimensional equation separating left from right-going waves can be
finite-differenced in space and time - pressures at spatial nodes are interpreted
as those at specific transducer locations - values available at different time
levels are interpreted as values stored in different computer locations.
Multilevel and multi-node schemes are easily developed which can be as
complicated as the need warrants. One such implementation is given by the
author in his U.S. Patent No. 5,969,638, “Multiple Transducer MWD Surface
Signal Processing,” awarded Oct. 19, 1999. However, the method is incomplete
in that the formulas terminate with the time derivative of the signal when it is
really its integral that is important. The model in Chapter 4 remedies this by
augmenting the basic approach with a highly robust integrator that successfully
handles the sudden starts and stops associated with short high-data-rate pulses.
When this surface signal processing method is used, all downgoing noise
regardless of shape and amplitude is virtually eliminated, allowing standpipe
mounted piezoelectric transducers to detect minute MWD signals accurately,
knowing only the local sound speed, which is separately measured or estimated.
We give examples of our success with the scheme. Recall that we had
previously considered the upcoming test signal
Search WWH ::
Custom Search