Signal Processor Design and Additional Noise Models - Measurement While Drilling Signal Analysis, Optimization and Design - page 175

6.8 Typical Frequency Spectra and MWD Signal Strength
Properties
We have discussed the physical properties of typical noise components in
the drilling channel. It is of interest, from an experimental perspective, to
examine typical MWD signals and how they might appear together with drilling
anomalies. Figure 6.8a displays “clean” frequency spectra for a siren pulser
operating at a constant frequency f at both low and high flow rates. Because the
sound generation mechanism and fluid-dynamical equations are nonlinear,
harmonics at 2 f , 3 f , 4 f and so on, will also be found, as shown. At the present
time, there is no clear method for their prediction and removal. Higher
harmonics are associated with inefficient signal generation and also complicate
surface signal processing. It is speculated that “swept” siren rotors might
minimize harmonic generation, but further study is required. The results in
Figure 6.8a were obtained in a wind tunnel and are “clean” in that they are free
of real noise.
On the other hand, the sketch in Figure 6.8b (from an unreferenced
Schlumberger source) shows a siren operating at 12 Hz in mud under typical
conditions. Note the existence of pump noise and mud motor noise. Because
mudpump noise propagates in a direction opposite to the upgoing MWD signal,
it can be effectively removed using directional multiple-transducer surface
signal processing techniques, e.g., as shown in Figures 4.4b,c. These would
eliminate the pump spectra in Figure 6.8b. On the other hand, mud motor noise,
which travels in the same direction as the upgoing MWD signal, would require
frequency-based filtering, effective only if the siren frequency were different
from that of the mud motor. In practice, MWD frequencies should always be
chosen so that they differ from that of the motor to facilitate noise removal.
Figure 6.8a. Signal strength harmonic distribution.