Advanced System Summary and Modern MWD Developments - Measurement While Drilling Signal Analysis, Optimization and Design - page 278

between 60 Hz and 0 Hz, that is, bringing the rotor to a complete stop with the
rotor fully open, using the frequency sequence “60 - 0 - 60 - 0 - 60 - 0 - 60 - 0
- 60 - 0.” Each “60 - 0” interval would take 6 cycles or 0.1 sec, so that 60
cycles are used per second. From the above, 0.00857 sec (or roughly, 0.01 sec)
is required to establish constructive interference and the 0.1 sec interval would
waste only 2 u 0.01 sec in noise tails at the beginning and end of each interval.
This leaves 0.08 sec of pure harmonic signal (or four wave cycles) for signal
identification. Thus, 10 bps is achievable as described; a higher rate is possible
if each frequency interval requires less than six wave cycles. The amplitude
pattern would be a wavelike with alternating bands with and without signal. The
60 Hz target carrier is doable, in practice, because 24 Hz is already realizable
from several service companies using the lead author's low-torque “rotor
downstream” designs.
What is 60 Hz in terms of siren rotation speed? If a rotor with N lobes
rotates at M rpm, then it will create MN cycles in 60 sec, that is, MN/60 cycles
per second. If N = 4 and 60 Hz is required, then M = 900 rpm is needed. This
may be demanding in terms of inertia and torque, since we have argued that the
rotor is brought to a complete stop between 60's. But we need not do this.
From Figures 10.2a,b,c, 40 Hz provides enough signal contrast to that at 60 Hz,
and we can consider alternatively the frequency sequence “60 - 40 - 60 - 40 -
60 - 40 - 60 - 40 - 60 - 40.” The 60 Hz would be associated with high
amplitude, owing to constructive interference, while the 40 Hz would be
associated with easily distinguishable waves of much smaller amplitude, owing
to destructive interference. Because the rotor is not brought to a full stop, a very
low torque siren may not be necessary, and mechanical inertia demands on the
drive motor would be reduced. Also, because time is saved by not completely
stopping, data rate can increase since more frequency cycles can be performed.
A siren with azimuthally wider rotors, which in the author's experience
produces larger 'p's, is associated with higher torques; because one does not
completely stop the rotor, the torque issue is now less of an issue.
It is important to emphasize the roles played by wave reflections. In a PSK
scheme where information is conveyed by introducing phase shifts to a carrier
wave, the downward wave from the pulser, upon reflection at the bit, adds to
later waves traveling uphole and introduces phase shift uncertainties associated
with ghost signals. The result is a type of randomness traveling up the drillpipe
that is not easily deconvolved. This issue is not discussed in the literature,
perhaps intentionally, but processing the downhole signal in any event adds to
signal processing demands. Even if multiple transducer surface signal
processing methods perfectly remove mudpump, desurger and uphole
reflections, the uncertainty created near the downhole pulser remains.
In theory, the processed signal must be further deconvolved to account for
wave interactions in the MWD collar. This is possible in principle, however, the
models available so far are crude and take the drillbit either as an open or solid