GB2312062A - Noise detection and suppression for wellbore signalling - Google Patents

Abstract

Methods and apparatus are described for identifying and removing noise from borehole telemetry or control signals. A spectrum of the signal and of an approximated noise-free spectrum of the signals are generated and compared so as to identify frequencies at which noise occurs. The approximated spectrum is preferably generated by an averaging or integrating process which essentially results in a smoothened variant of the signal spectrum. The comparison includes dividing the signal spectrum by the approximated spectrum, and setting a threshold in the resulting dimensionless peak spectrum. The remaining parts of the peak spectrum are identified as noise. The dimensionless spectrum is found to have major advantages when having to identify noise frequencies.

Description

Noise Detection and Suppression System for Wellbore Telemetry The present invention relates to telemetric systems communicating data through noisy transmission channels. The present invention particularly relates to telemetric systems in formation evaluation or borehole telemetry. More specifically it relates to the identification and suppression of noise in such channels.

BACKGROUND AND OBJECT OF THE INVENTION In the development, completion, and operation of natural hydrocarbon reservoirs, various telemetric systems and techniques are known and employed to achieve what is known in the art as measurement while drilling (MWD).

For the purpose of this application, MWD includes any type of data transmission from sensor units in the drill bit, bottom hole assembly, or any other part of the sub-surface drill string. Another acronym often encountered in the art besides MWD is LWD (Logging While Drilling). MWD includes in particular low data bit rate transmission systems, as operating below 10 kHz, preferably below 1 kHz, such as acoustic telemetry through the drill string itself, or mud pulse telemetry.

In the latter, currently prevailing technique data are transmitted by means of a mud pressure pulse generator located inside the drill string. The system generates pressure pulses in the drilling fluid or mud, typically by way of a valve or siren type of device. The pulses are detected at the surface by suitable means, e.g., pressure sensors, strain gages, accelerometers, and the like, which means are in general directly attached to the drill string.

borehole or wellbore telemetry is a well established technology.

Improvements to this technology as have been made over the past decades are published for example in a large number of patents, including the US Patents US-A-3 790 930, US-A-3 820 063, US-A-4 739 325, US-A-4 932 005.

Of particular interest for the scope of the present invention are the numerous attempts being made to improve the data detection of the transmitted data at the surface. It should be noted that the drilling process presents an exceedingly noisy environment for telemetry owing to the mechanical generation of broadband noise and to the drilling fluid circulation system.

To improve the signal-to-noise ratio, the data as gathered by the sensor units can be encoded such that the distortion by noise has less impact on the data recovery. Usually employed encoding schemes include Frequency Shift Keying (FSK), Phase Shift Keying (PSK) or m-ary pulse coding. Alternatively a binary non return to zero coding may be used. Different encoding methods are described for example in US-A-3 789 355 or US-A-4 562 559.

In US-A-5 381 092 the signals from those sensors which evaluate the earth formation are subdivided prior to transmission into a plurality of groups, each group represented by one value.

In US-A-5 055 837 an attempt is described to improve the quality of the transmission by determining a transfer function which characterizes the transmission properties of the drilling fluid column in the drill pipe.

In an acoustic telemetry system, as described in US-A-5 128 901, the data signals are (pre-)conditioned to counteract distortions caused by the drill string.

A filtering technique to cancel or minimize noise in the transmitted data signals is disclosed in US-A-4 878 206. This known approach uses independent measurements of the vibrations of the drill string at the surface to remove pressure disturbance caused by these vibrations and affecting the mud column pressure. A similar technique is known from US-A-5 289 354.

Probably the closest prior art is an undocumented, manual approach to noise filtering, wherein an experienced operator searches the frequency spectrum of the transmitted signal for obvious peaks, marks those frequencies at which the peaks occurs and selects appropriate filters to remove them from the signal.

This procedure depends on the quality of the operator. It can not be used to provide a deterministic, reliable, and real-time operation service.

A straightforward automation of the above process faces difficulties arising from the nature of wellbore telemetry signals. Those signals are characterized by a low S/N ratio, large fluctuations in the absolute value of the signal and drifts or baseline shifts.

In view of the above cited prior art it is an object of the invention to provide an automated telemetry system with improved noise reduction and cancellation. The system should be compatible with or independent from the various transmission media and encoding methods. It is a particular object of the invention to provide such a system for downhole telemetry in the low frequency domain, in particular for mud pulse telemetry.

SUMMARY OF THE INVENTION The above-mentioned objects of the invention are achieved by apparatus and methods as set forth in the appended claims.

It is seen as a first important feature of the invention that noise frequencies are identified from a dimensionless spectrum, which is in a preferred embodiment of the invention generated by dividing the signal spectrum and another spectrum representing an approximation of the noise-free signal.

The spectrum of the gathered telemetry signal and an approximation of a noise-free spectrum of the gathered signal is generated, preferably using a Fourier-type transformation. The approximated spectrum is preferably generated by averaging or integrating the signal spectrum. This process is based on the observation that noise in downhole telemetry often has a narrow bandwidth.

Dimensionless is defined for the scope of the present invention as having the nature of a pure number, with a value independent from the choice of units. The property of being dimensionless should not be confused with the numbers having arbitrary units such as often encountered in cases where the absolute value in SI or other internationally standardized units is of no importance for the result displayed.

From the dimensionless peak spectrum noise frequencies can be determined by, for example, setting a threshold above which every part of the peak spectrum is regarded as noise. In this process, it is an important advantage for borehole telemetry to be independent of absolute values, as the telemetry signals in this technical field are prone to various drifts and sudden changes caused by external sources.

To facilitate computation, the signal is prior processing subdivided into adjoining time intervals.

Obviously the invention is not restricted to the transmission of signals to the surface, but could also be employed in downhole receivers which are responsive to control commands sent from the surface.

These and other features of the invention, preferred embodiments and variants thereof, and advantages will become appreciated and understood be those skilled in the art from the detailed description and drawings following hereinbelow.

BRIEF DESCRIPTION OF DRAWINGS FIG. 1 shows a schematic view of a mud pressure pulse generator and drill string as providing the background for an example of the present invention; FIG. 2 shows a flow diagram illustrating major functions of a noise identification and removing system in accordance with an example of the present invention; FIG. 3 shows a flow diagram illustrating major functions of a noise identification step within the system of FIG. 2; FIGs. 4A-4C show spectra illustrating major aspects of an example in accordance with the invention.

MODE(S) FOR CARRYING OUT THE INVENTION Referring now to the drawings, there is shown in FIG. 1 a tubular MWD tool 10 connected in a tubular drill string 11 having a rotary drill bit 12 coupled to the end thereof and arranged for drilling a borehole 13 through earth formations 14.

As the drill string 11 is rotated by the drilling rig, substantial volumes of drilling fluid ("drilling mud") are continuously pumped by mud pumps 15 down through the drill string 11 and discharged from the bit 14 to cool and lubricate the bit and carry away cuttings removed by the bit. The mud is returned to the surface along the annular space 16 existing between the walls of the borehole 13 and the exterior of the drill string 11. This circulating stream of mud can be used for the transmission of pressure pulse signal from the MWD tool 10 to the surface.

The MWD tool 10 of this example is an integral part of the drill-string bottom hole assembly. It comprises measuring devices 101 for environmental and drilling parameters and appropriate encoders 102 to reduce and refine electrical signals representative of the measured parameters for transmission via mud pulse telemetry signals to the surface. In this example the MWD tool measures direction and inclination of the hole, gamma radiation, temperature, and weight and torque on bit. Sensors and tools for other parameters such as downhole pressure, downhole resistivity or conductivity of the drilling mud or formation, neutron spectroscopy etc. might be added. It should however be obvious that the present invention is not concerned with any specific kind of parameter or measuring device as used in the wellbore.

Electrical power for the operation of the tool is provided by battery producing electrical energy. The tool 10 also includes a modulator, or mud siren, 103 which selectively interrupts or obstructs the flow of the drilling mud through the drill string in order to produce pressure pulses in the mud. Suitable generators are for example described in US-A-4 785 300, US-A-4 847 815, US-A-4 825 421, US-A-4 839 870 or US-A-5 073 877.

The modulator 103 is controlled such that the pressure pulses are produced in the form of encoded acoustic data signals which correspond to the encoded signals from the measuring devices 101. These signals, typically in the form of binary coded sequences, are transmitted to the surface by way of the mud flowing in the drill string.

In the present example NRZ (Non-Return-to-Zero) telemetry is used to communicate information to the surface. In NRZ modulation the symbols are binary ones and zeros. The system states are the modulator closed (corresponding to a one) and the modulator open (corresponding to a zero). Thus, if two succeeding bits are the same the modulator does not move. If a one follows a zero the modulator closes, if a zero follows a one the modulator opens.

Other signal modulation techniques are usable, and selection of the specific encoding and modulation schemes to be employed in connection with the operation of the modulator are matters of choice. A number of possible modulation schemes for acoustic borehole telemetry are described by S.P.Monroe, "Applying digital data-encoding techniques to mud pulse telemetry", Proceedings of the 5th SPE Petroleum Computer Conference, Denver, 25th-28th June 1990, SPE 20236, pp. 7-16.

When these signals reach the surface, they are detected, decoded and converted into meaningful data by a suitable signal detector 36, which in the present example includes a electro-mechanical transducer which is generally known in the art as SPT (Standpipe Pressure transducer) 17. Transducers suitable for a acoustic signal/pressure conversion into electrical signals are also found in the published UK Patent GB-A-2 140 599, in US-A-5 222 049, or in the published International Patent Application WO-A-95/14 845.

The analog signal of the SPT is sampled at an appropriate frequency to derive a digitally coded representation of the analog signal, which then can be further processed as described in the following.

Referring now to FIG. 2, the major steps for removing the noise from transmitted and received signals are described.

The first step 21 comprises the identification of noise by determining its frequency spectrum or at least its major frequency component or components in a time interval, hereinafter also referred to as the data vector, which taken from the received signal.

In a second 22 and third step 23 noise frequencies are selected and removed from the signal.

After noise removal, an adjacent time interval, i.e., the subsequent data vector of the received signal is selected (step 24). And the procedure is reiterated until the complete signal is processed.

It should be noted that the above-described sequence of steps not necessarily reflects a fixed temporal order. It is for example feasible to process several intervals of the signals in parallel and thus accelerating the noise identification.

The noise identification is regarded as one of the key steps of the present invention; it is described in greater detail making reference to FIG.3.

The noise identification includes the step 31 of selecting a data vector. In the current example the data vector spread over 60 to 120 seconds of the received signal. It is centered around a time reference point.

The following step 32 comprises several processes: As the signal, and hence the data vector, may be subject to slow trends or drifts, methods and means for removing these trends are employed. Preferable this step is performed before any transformation into the frequency domain. Normally the best-fit line using least-square based is sufficient to remove a trend from the data. Other means and methods are readily available to a person skilled in the art. The parameters for the line fit are retained for a reconstruction of the data after the noise removal.

The data vector is then transformed by a Fourier transformation into the frequency domain. In a particularly preferred variant of the present example the chosen data vector is multiplied by a suitable windowing function (such as a flat window with cosine end-tapers), and padded with zeros - lengthening the data four or eight fold, thereby enhancing the resolution of the spectrum.

Zero padding introduces side-bands next to the noise peaks, but these are removed in later step 36 as is described below. The Fourier transform itself can be performed for example by a accordingly programmed multi-purpose microprocessor, a digital signal processor (DSP), or an application specific integrated circuit (ASIC).

After the spectrum of the data vector is determined in absolute values, in one variant of the invention, the spectrum is scaled such that if only signal were present the spectrum would be approximately flat. For the NRZ telemetry of this example the spectrum is roughly proportional to the inverse of the frequency, so the spectrum is multiplied by the frequency to flatten it. This step 33 is not essential, but can improve results.

In a further step 34, a running median filtering process is applied to the spectrum. The filter has a window length of 0.2 Hz. Essentially this step results in a smoothened version of the original spectrum, in which all peaks are largely reduced in height. Though this method generally gives the best results, other methods, which can be summarized as a weighted median or mean value based, could be applied to generates a smoothened version of the spectrum of the signal. Other integrating methods are feasible.

In another example the (non-windowed) median of the complete spectrum is used as an approximation of the noise-free spectrum.

In yet a further example, a windowed or shifting mean of the spectrum is used as an approximation of the noise-free spectrum.

From these different variants, the windowed median filtered spectrum provides the best results under usual circumstances.

Returning now to Fig. 3, in the next step 35 the smoothened spectrum is used to identify the peaks in the original spectrum.

This is done by dividing both spectrums and using a threshold to suppress all amplitude below a threshold, which is 5 in the present example. The advantage of using a division rather than the difference spectrum lies in that the resulting spectrum is independent from absolute values or units.

To eliminate those side bands which are introduced by the zeropadding in step 32, from the resulting peak spectrum, any nonzero intervals in the peak spectrum that are close to, but have a maximum value smaller than, another non-zero interval are discarded. Now the frequencies within each interval at which the peak spectrum attains a maximum are found. If the maximum is above a second threshold value (usually chosen to be greater than 6), then this frequency is identified as being narrow-band noise frequency. If the maximum is between the two thresholds then the frequency is only identified as a noise frequency if it is the same as or very close to a frequency that has previously been identified as a noise frequency. This is to allow for frequency drift, and for noise sources whose amplitude goes up and down.

The results of the above described operations are displayed in Fig. 4. The frequency scaled spectrum with the prominent noise frequencies (a,b) is shown in Fig. 4A. A smoothened version of this spectrum is produced by windowed median filtering (Fig.

4B).

In Fig. 4C the result of a division of the original spectrum (Fig. 4A) and its median filtered version (Fig. 4B) is shown.

The resulting "peak spectrum" is dimensionless. The threshold separates the (assumed) noise (a,b) from the signal.

The result of the process is a (possibly empty) list of frequencies at which there is narrow-band noise.

The noise removal process make use of the frequency identified through the previous steps as contaminated with noise. With these information a variety of different filtering techniques can be applied to the original data to reduce or cancel the noise in the signal.

One possibility is to employ notch filters with notches at the noise frequencies, or adaptive Least-Mean-Square (LMS) methods described as such by B. Widrow and S.D. Stearns in: "Adaptive Signal Processing", Prentice Hall, 1985, on pp. 212-224 and pp.

99-116, respectively.

The method used for removing the noise frequencies from the telemetry signal in the present example includes the step of generating, for each of the identified noise frequencies, two noise vectors of mono-chromatic noise at that frequency with a 90 degree phase difference between them, e.g., if the noise frequency is F then the two noise vectors could be [1] N1 =sin(2ltFt) N2 =cos(2?tFt), with t from a time interval which is much shorter than the original interval, in this example 12 seconds as compared to the 60 to 120 seconds length of the noise identification vector.

Synthesizing noise for all the noise frequencies produces a noise matrix N. Assuming that the data D is corrupted by an additive noise component according to [2] D =AN+S1 where D is the vector of data, S is the underlying signal, N is the matrix of noise synthesized at the narrow-band noise frequencies, and A is the vector of (unknown) noise amplitudes.

The vector A is found by minimized the residual E, given by

The solution to eq. [3] is given by T T 1 [4] A = (D N ) (N N where T denotes the transposed matrix, and the signal is given by reordering equation [2]: [5] S = D A N.

This process may also be carried out over a slightly longer noise removal vector, multiplied by a windowing function that is one over the original range of the removal vector, as defined above, and beyond goes smoothly to zero, in which case the noise vectors are multiplied by the same windowing function.

The signal has its trend reinstated, adding back the linear function removed during noise identification, and a data jump is avoided by adding a constant to the data such that the last point of the previously-processed data and the first point of the current data are the same.

The reference time is then moved forward by the length of the noise-removal vector, and the process is repeated.

An improvement can be made to the method described above when the signal is also subject to time-domain spike noise, and the data has a wider bandwidth than the signal alone. Spike noise causes problems with Fourier techniques, since when a large spike is within the noise-identification vector, the frequency transform is dominated by the spike. Spike removal may be achieved by using a running-median filter on the data, and comparing the original data with the median filtered version. An example of this method is described as such in "Running median filters and a general despikker", in: Bull. Seism. Soc. Am. 72 (1982), pp. 331-338, by J.R. Evans. If the difference is greater than a threshold, the original data is replaced by the median filtered version - if not then it is unaltered. To this end, the data is low-pass filtered with a bandwidth that exceeds the signal bandwidth, preferably by a factor of two. The median filter length is then chosen so that the spikes, after low-pass filtering will be contained within the filter length. A suitable length L is given by

where F, is the Nyquist frequency and FL is the frequency at which the data has been low-pass filtered.

The process is performed on a general purpose computing machine which receives the signal from the SPT and on which a suitable program, e.g. MATLAB is installed. Other possibilities for implementing the method include the use of a dedicated microprocessor, known as such as DSPs or ASICs.

Claims (22)

1. Noise filtering apparatus for wellbore telemetry signals, comprising means for identifying a frequency or a range of frequency of noise in a telemetry signal, characterized in that said identifying means include means for generating a dimensionless spectrum of said signal.

2. The receiver of claim 1, wherein the means for generating a dimensionless spectrum comprises transform means for generating a signal spectrum of at least a part of said signal, means for generating an approximation of a noise-free spectrum of said at least a part of said signal, and means for dividing said signal spectrum and said approximated spectrum.

3. The receiver in accordance with claim 2, wherein the means for generating the approximated spectrum comprises a weighted means based filter.

4. The receiver in accordance with claim 2, wherein the means for generating the approximated spectrum comprises a median filter.

5. The receiver in accordance with claim 1, further comprising means for filtering said identified frequencies from the signal, said filtering means being responsive to the means for identifying noise frequencies.

6. The receiver in accordance with claim 1, further comprising means for removing spike-type noise from the telemetry signal prior to the generation of a dimensionless spectrum of said signal spectrum.

7. A receiver in accordance with claim 1 and any combination of claims 2 to 6.

8. Use of a receiver in accordance with claim 1 for transmitting data related to subsurface conditions or control signals for downhole apparatus.

9. Transmission apparatus for gathering data related to subsurface conditions and for transmitting said data to the surface, said apparatus including first transducing means in a well for converting gathered data into signals to be in operation transmitted to said surface, second transducer means for reconverting said signals into prosecutable data, processing means for processing said data, characterized in that said processing means comprises transform means for generating a signal spectrum of at least a part of the signal, means for generating an approximation of a noise-free spectrum of said at least a part of said signal, means for dividing said signal spectrum and said approximated spectrum so as to identify noise frequencies.

10.Method for identifying a frequency or range of frequency of noise in a wellbore telemetry signal, comprising the step of generating a dimensionless spectrum of said signal, from which said noise frequency or frequencies are identified.

1l.The method of claim 10, comprising the steps of - transforming at least a part of said signal to transform generate a spectrum of said at least a part of said signal; - generating an approximation of a noise-free spectrum of said part of the signal; - generating a dimensionless spectrum by dividing said signal spectrum and said approximated spectrum.

12.The method of claim 11, wherein the step of generating an approximated spectrum comprises the step of weighted means based filtering of the signal spectrum.

13.The method of claim 11, wherein the step of generating an approximated spectrum comprises the step of median based filtering of the signal spectrum.

14.The method of claim 10, wherein the step of identifying the noise frequencies includes the step discarding all values of the dimensionless spectrum which have a value below a predetermined threshold.

15.The method of claim 11, wherein the step of generating the signal spectrum comprises a Fourier transformation or a mathematical equivalent thereof.

16.The method of claim 10, wherein prior to the step of generating the signal spectrum trends in said signal are removed.

17.The method of claim 10, wherein prior to the step of generating the signal spectrum spikes in said signal are removed.

18.The method of claim 11, wherein prior to the step of generating the signal spectrum a zero signal is added to both sides of the at least a part of the signal.

l9.The method of claim 11, wherein prior to the step of generating the signal spectrum the at least a part of the signal is multiplied by a window function with smooth edges.

20.The method of claim 10, further comprising the step of removing noise from the signal spectrum by suppressing noise frequencies as identified in the previous step.

21.The method of claim 20, wherein the noise frequencies are suppressed by using a noise matrix N derived from two linear independent state vectors for each identified noise frequency, and minimizing the residual between the gathered signal and a noise vector derived from multiplying said noise matrix N with a vector of unknown noise amplitudes.

22.The method of claim 10 with any combination of claims 11 to 21.