NO162687B - PROCEDURE FOR TELEMETERATION OF DATA IN A DRILL HOLE FILLED WITH DRILL. - Google Patents

Description

This invention relates to a method for telemetering data in a borehole filled with drilling fluid, comprising the generation of individual data-carrying fluid pressure signals down in the borehole. These signals are arranged in a coded sequence, in the fluid and this coded sequence represents the size of a parameter down in the borehole.

It has long been known within the oil industry to obtain data from drilling operations in a borehole which provides valuable information for the drilling crew. Information such as the weight of the drill bit, inclination and direction of the drill hole, condition of the drill bit, fluid pressure and temperature at the bottom of the drill hole and radioactivity of substances that surround or are processed by the drill bit can be expressed in quantities that are of interest to the drilling crew. A number of previous proposals for measuring such values during drilling and transferring them to the ground surface are available, e.g. in U.S. Patent Nos. 2,787,795, 2,887,298, 4,078,620, 4,001,733, 3,964,556, 3,983,948 and 3,791,043.

The most promising of these proposals in practical terms is the transmission of signals by means of pressure pulses in the drilling fluid. Various methods have been proposed for producing such mud pulses either by controlled throttling of the mud flow using a throttle valve between the mud flow inside the drill stem (high pressure side) and the annular space around the drill stem (low pressure side).

It is further proposed to produce mud pressure pulses by means of valves which either throttle the mud flow inside the drill stem or divert some of the flow to the low-pressure zone in the annular space around the drill stem. Such valves are necessarily slow because when they are used inside the drill stem the valve must control very large quantities of mud and when they are used to control a diversion, the valve is also necessarily a slow motorized valve as a result of the very large pressure difference. Such a motorized valve as e.g. is placed between the inside of the drill stem and the annular space, emits, as a result of a measurement down the borehole, slow decreases and slow increases in the mud pressure. These are later detected on the earth's surface.

In order to better understand the operation of a slow-acting motorized valve which has previously been proposed, is shown in fig. IA and whose opening and closing is a function of time. There, the abscissa is a measure of the time t, while the ordinate represents the degree of opening of the valve R.

where Sq is the total area of the opening and S(t) is the area that is open in time t during opening or closing of the valve. R=0 means that the valve is closed and R=l means that the valve is fully open. The times included in operating the valve are the following: t a ^ <=> OAt -L the time when opening begins tj^<v>^ = OB^ the time when the valve is fully open t ^ = OC-^ the time when closing begins

t^V^ = OD^ the time when the valve is completely closed.

The time interval:

denotes the time for opening or closing the valve. The time interval: denotes the time for open valve. The total operating time is thus:

In one of the above proposals, Ta^ = 1 second, Tj_^<v>^ = 2 seconds, and Tfc ^ = 4 seconds. These relatively slow opening and closing times for the valve produce correspondingly slow decreases and increases in the mud pressure on the soil surface (see fig. IB).

It appears here that the mud pressure decreases from its normal value of e.g. 70 kg/cm o when the valve is closed to the lowest value 52.5 kg/cm 2 when the valve is closed. The times for these pressure variations are as follows:

(s)

t^a = OE^ is the time when the mud pressure begins to decrease from its normal level of 70kg/cm^.

(s)

t^b = OF^ is the time when the mud pressure has its lowest value 52.5kg/cm^, and this pressure is maintained until the time

t<<;>' - and,.

(s)

t^c = OG^ is the time when the mud pressure starts to increase, ,

(s)

_t^d = OH^ is the time when the mud pressure regains its normal level of 70kg/cm^. The pressure thus decreases in the time interval Tx(<s>) = tj^ - t^- , remains constant in the time interval

(s) (s) (s)

T_v ' = t, ' - t,Z and increases from the reduced value to

2 lc lb 3 , , , . , .

(s) (s) (s) normal value in the time interval T- = t, •/ - t, . It

3 ld lc

total time that the sludge flow takes to pass the valve with a single operation of the valve is therefore:

The values in fig. IA such as

t a (v), t d <<V>), t c (v),t Q <v),T a (v),T, d (<v>) and Tt.i(v) are marked with v to indicate that these values apply to the value below the earth's surface. On the other hand, the values t(s) (s) (s) (s) (s) (s) (s) (s) , f tla ' fclb ' tlc ' tld ,L1 ' l2 ' L3 and rt pa tig ' IB denoted by s to indicate that these values apply to measurements on the earth's surface. This distinction between the values denoted by v and s is important to fully understand the new features that the invention represents. In this connection, it is important to distinguish between cause and effect, or between phenomena that occur down the borehole

near the valve and those detected on the soil surface.

An important feature of the previous proposals is

based on the conditions

These conditions show that the time for the decrease or increase of the pressure on the earth's surface is the same as the corresponding time for the opening and closing of the valve and the time during which the pressure is essentially constant at a reduced level is the same as the time when the valve is fully open. In other words, a decrease with a subsequent increase in the mud pressure on the soil surface will exactly correspond to the opening and closing of the valve. These conditions expressed by equations (6), (7) and (8) shall below be referred to as times for slow variations in the pressure.

The time for slow pressure variation as proposed earlier is not suitable for remote measurement during drilling, especially when several parameters down the borehole are to be measured. At the time when the first parameter has been measured, coded and transferred to the earth's surface and there decoded, the borehole may have become deeper and the next parameter is no longer available for measurement. Relatively long intervals are necessary for the transformation of measured data into a form suitable for detection and recording. The entire monitoring is prolonged and time-consuming. Furthermore, various interfering effects such as the pulsations caused by mud pumps and noise in connection with various drilling operations cause additional difficulties. A slow-acting motorized valve as proposed earlier is not suitable to satisfy today's requirements for accuracy.

One of the essential aspects of the present invention is the use of hydraulic shock waves for remote measurement during drilling. These shock waves are produced by very fast, in practice almost instantaneous operation of a valve between the inside of the drill stem and the annular space around the drill stem. When the valve is opened suddenly, the pressure in the immediate vicinity of the valve will drop and return to normal almost instantly and a sharp negative pressure pulse is produced, and on the other hand when the valve is suddenly closed, a sharp positive pulse is produced. The elasticity of the mud column is used to support the generation and transmission of such shock waves. The phenomenon is analogous to the well-known water hammer effect previously used in hydraulic transmission systems known from John Parmakian on "Water Hammer Analysis", Prentice Hall, Inc., New York, N.Y. 1955 or V.L.Streeter and E.B.Wylie on "Hydraulic Transients" McGraw-Hill Book Co., New York, N.Y.

Important features of the invention such as the generation and detection of hydraulic shocks appear in fig. 2A and 2B. Fig. 2A shows the opening and closing of a quick-acting valve which produces a shock wave while fig. 2B shows pressure variations detected at the earth's surface as a result of the operation of the valve. Symbols such as A1, B1, C1, D1, ta , tfa , tc ,, Tfl , Tfa and Tfc in fig. 2A has the same meaning as corresponding symbols in fig.

IA. The time scales in fig. 1A, 1B, 2A and 2B are substantially distorted for ease of description and for clarity of explanation.

The first thing to note when considering fig. 2A is that the times for opening and closing the valve are several orders of magnitude shorter than corresponding times than in fig. IA. The time T 3.^ is in fig. In the 1st second, while the one in fig. 2A is 5 milliseconds. The same is the case when it comes to the time for opening the valve. tk is in Fig. IA equal to 2 seconds, while in Fig. 2A is 100 milliseconds.

For all practical purposes the opening and closing can therefore

of the valve according to fig. 2A is considered instantaneous or nearly instantaneous.

Rapid or almost instantaneous opening and closing of the valve has an important and more far-reaching impact on the remote measurement system when measuring during drilling. The pressure variations detected on the earth's surface according to fig. 2B shows no resemblance to the pressure variations obtained in FIG. IB. Equations (6), (7) and (8) have been pointed out above, which show the relationship between the effects on fig. IA and IB. However, analogous conditions do not occur with the effects on fig. 2A and 2B.

As can be seen from fig. IA and IB, opening the valve will result in a corresponding reduction in the mud pressure on the soil surface and, on the other hand, closing the valve will produce a corresponding increase in pressure.

It should be emphasized here that, with the known technique, opening the valve produces a single reaction, namely a decrease in pressure and a subsequent closing of the valve produces another simple reaction, namely an increase in pressure. On the other hand, the rapid opening of the valve according to the invention as shown in fig. 2A two reactions, namely a rapid decrease and a subsequent increase in the pressure in the form of a negative pulse "M" as shown in fig. 2B. This is in contrast to the case in fig. IA and IB where an opening and subsequent closing of the valve is necessary to produce a reduction and subsequent increase in pressure. A quick closing of the valve as shown in fig. 2A, an increase and subsequent decrease in mud pressure produces a positive pulse "N" as shown in FIG. 2B. Such an increase and subsequent decrease in pressure does not occur with the previously known technique. With the invention, two shock waves are produced by a single operation of the valve. A waveform as shown in fig. 2B, which comprises a negative and a positive pulse, shall be referred to below as a valve wave. The pressure pulses in connection with a valve wave have a start of several hundred kg/cm 2/sec. and has a very short duration.

It should also be pointed out that it is the speed

of the phenomena occurring at the valve wave that are of interest. The times in fig. 2B is the following!

(s)

= OK is the time of appearance of the negative pulse

"M",

(s)

t2 = OL is the time of the delay of the negative pulse "M",

(s)

t^ = OM is the time of appearance of the positive pulse "N",

(s)

t^ = ON is the time of delay of the positive pulse "N".

(s)

The time Tn represents the length of the negative pulse "M"

(or the positive pulse "N") is 100 milliseconds while

(s)

the time Tm from the appearance of the negative pulse "M" to the appearance of the positive pulse "N" is 110 milliseconds. The total time for the current shown in fig. 2B is:

i.e. 210 milliseconds while the total time for the current shown in fig. IB is 4 seconds (see equation (5).

The curves in fig. 1A, 1B, 2A and 2B are simplified and idealized to eliminate unimportant effects. It should be noted that in fig. 2B, the valve is at least partially open during the time interval t^ (s) to t^ (s). During this time, there is a small pressure drop that is eliminated at the detection site by means of a suitable filter. Such a pressure drop is not shown in fig. 2B.

It should also be pointed out that the numerical values in fig. 2A and 23 are only given as examples and mean no limitation of the invention.

The effect which is explained in connection with fig. 2A and 2B are designated as the time of hydraulic shock waves. There is thus a clear difference between the time for the hydraulic shock waves in fig. 2A and 2B and the time for the slow pressure variations in fig. IA and IB.

By providing a time for hydraulic shock waves, a remote measurement system is achieved that can transmit a high degree of information per unit of time. Such a system is significantly better suited to satisfy today's requirements or a system that is based on time for slow pressure variations.

When the present method makes use of a valve, this is operated under the control of one or more sensors which sense one or more parameters down in the drill hole near the drill bit. A single measurement of each parameter is represented by a series of valve waves. Each valve wave corresponds to a single opening and closing of the valve.

The sequence of the valve wave representing a usable signal will, when detected at the earth's surface, usually be mixed with various interfering signals, such as are produced by operation of the mud pump and other drilling operations. A large pump placed on the ground surface is used to pump drilling fluid down through the drill stem and drill bit and back to the ground surface at the annular space between the drill stem and the borehole wall. According to the invention, the interfering effects caused by the pump are eliminated by a process which takes into account the periodicity of these effects. Other influences in connection with the drilling operations usually appear as a noise signal that includes a relatively broad frequency spectrum. This noise signal is in some cases white noise and in other cases far from white noise. A digital filtering system which can be matched filters or pulse-shaped filters or a peak filter is used to remove the noise signal. The matching filter increases the signal-to-noise ratio at the receiver site, a pulse-shaping filter reduces the square of the difference between the desired output signal and the actual output signal, while the peak filter transforms the useful signal by compressing it into a sufficiently sharp signal so that it can be distinguished from the background noise. A special technique is used to adapt these filters to the purpose of the invention. Such a technique requires the storage and subsequent reproduction of two reference signals. The first reference signal is a valve wave produced by opening and closing the valve and the second reference signal represents noise caused by the drilling operations. Detection and storage of the first reference signal is achieved by removing weight from the drill bit and stopping the drilling but maintaining the mud pump's normal operation. A signal that is free of ambient noise is thereby obtained. Detection and storage of the second reference signal is achieved during drilling during a period of time where the valve is closed. A suitable digital computer system receives data representing one or both of these reference signals and derives from the data a memory function for matching filters, for pulse shaping filters or peak filters.

In addition to the previously mentioned known technology, the following should be mentioned: U.S. patent 3,716,830 describes a device for monitoring the condition of a borehole during drilling, with the production of pressure changes in a mud circulation system.

Norsk Utlegningsskrift 151,907 shows a system for monitoring drilling operations during drilling, where pressure changes are produced in the circulation system with the help of a control device.

Norsk Utlegningskrift 140,648 is probably somewhat more of interest as it describes a method for receiving data signals that have been transmitted over a transmission channel with disturbing noise, including processing with a view to correcting the masking effect of the noise on the data signals.

Based on the known technology, the initially stated method is based on the assumption that the individual data-carrying fluid pressure signals down in the borehole lead to the generation of corresponding individual data-carrying pressure changes near the surface, as the borehole fluid is also affected by disturbing pressure changes caused by the drilling operations and which thereby mask the aforementioned data-bearing pressure changes. The pressure changes in the drilling fluid representing the superposition of the aforementioned data-carrying pressure changes and the aforementioned disruptive pressure changes are received near the surface, and the superimposed pressure changes are converted into electrical superimposed digital signals.

The new and distinctive features of the procedure according to . the invention is specified in more detail in the patent claims.

The invention will be explained below in more detail in the form of exemplary embodiments with reference to the drawings. Fig. IA shows a curve for operating a slow-acting valve which is used in known technique. Fig. 1B shows pressure variations which are detected on the earth's surface as a result of the valve operation in fig. IA. Fig. 2A shows operation of a quick-acting valve that can be used in the method according to the invention. Fig. 2B shows pressure variations detected on the surface as a result of the valve operation in fig. 2A. Fig. 3 schematically shows a drilling equipment for simultaneous drilling and measurement. Fig. 4A schematically shows part of the equipment down in the borehole, in particular the remote measurement equipment. Fig. 4B schematically shows part of the equipment in fig. 4A. Fig. 5A shows a block diagram of the electronic processing equipment shown within dashed lines in Fig. 4A. Fig. 5B shows a block diagram of the power source including a capacitor charge-discharge device that provides sufficient energy to operate the valve in the remote measurement equipment. Fig. 5C shows a block diagram of the electronic equipment used to automatically disconnect the drive energy for the valve. Figs. 5D and 5E show curves to facilitate the explanation of the automatic cut-off of the operating energy for the valve. Fig. 6A, 6B and 6C schematically show the operation of the hydraulic automatic disconnection device for the valve. Fig. 6D shows a detailed embodiment of the device shown in fig. 6A, 6B and 6C. Fig. 6E shows a block diagram of the electronic safety device for the valve. Fig. 7A shows the attachment and housing for the remote measurement equipment. Fig. 7B shows the cross-sectional shape of the centering device in fig. 7A. Fig. 7C shows in perspective the application of the connection for the housing part 250b of fig. 7A. Fig. 8A to 8E show curves representing variations in the pressure measured on the earth's surface and the corresponding values for T ^ (the time for opening or closing the valve) and for (the time for open flow). The curves show the results of certain tests carried out with the intention of obtaining optimal conditions for hydraulic shock waves, where Fig. 8F corresponds to an exact reproduction of the pressure signal with a valve wave received at the earth's surface from a depth of 3000 meters in a real borehole. Fig. 9 shows a block diagram of the equipment on the surface of the earth to process pressure signals from down the borehole and consists of an adaptive filter for eliminating ambient noise when this is white noise. Figs. 10A to 10G show curves of particular waveforms and pulses as a function of time to facilitate explanation of the operation of the equipment of Figs. 9. The time axes in fig. 10A to 10C and the time axes of Figs. 10D to fig. 10G are arranged below each other so that the signals and waveforms can be compared with each other. Fig. 10A comprises three curves relating to three components of the signal which is detected on the earth's surface, and which represents an information-carrying signal, the pump noise from one or more pumps, and ambient noise. Fig. 10B contains three curves, namely the delayed information-carrying signal, the delayed pump noise and the delayed ambient noise, which delay Tp represents a period of the pump function (when several pumps are used, the pressure variations, if not sinusoidal, are nevertheless periodic because their operation is mainly in phase) . Fig. 10C shows two curves corresponding to the difference of corresponding curves in fig. 10A and 10B. One of these curves represents ambient noise while the other represents the information-carrying signal. Fig. 10D shows a curve representing the output signal from a digital filter or a cross-correlator according to the invention. This function is essentially similar to the signal in fig. 10C. The digital filter used here can be an adaptation filter, a pulse shaping filter or a peak filter.

Fig. 10E shows a curve corresponding to fig. 10D

but delayed a suitable time.

Fig. 10F shows a curve similar to fig. 10E

but reversed in time.

Fig. 10G shows the result of the comparison

of fig. 10D and 10F and represent times corresponding to the pulses that appear simultaneously in these figures.

Fig. 11 schematically shows certain operations i

a digital filter.

Fig. 12 shows a block diagram for a device

for storing an information-carrying signal or a noise signal.

Fig. 13 shows a block diagram for part of the equipment on the earth's surface, including a correlator for noise elimination. Fig. 14 schematically shows a block diagram for part of the equipment on the ground surface including an adaptation filter for noise elimination when the noise is not white noise. Fig. 15 shows a block diagram for some of the equipment

on the Earth's surface, including a pulse shaping filter.

Fig. 16 schematically shows certain operations carried out in a pulse-shaped filter. Fig. 17 shows a block diagram for part of the equipment on the earth's surface including a peak filter where the peak filter transforms a double valve wave into a corresponding pair of peaks. Figs. 18A to 18F show six possible choices of peak positions for a pair of peaks produced by means of the device of fig. 17. Fig. 19 shows a block diagram of part of the equipment on the ground surface, with a peak filter used to transform a simple valve wave into a corresponding simple peak. Fig. 20A to fig. 20F shows six possible choices of peak positions for a single peak produced by means of the device of fig. 19. Fig. 21A to fig. 21C schematically shows certain operations in connection with a peak filter for different delay times. Fig. 21A corresponds to a desired peak with time index 0. Fig. 21B corresponds to a desired peak with time index 1. Fig. 21C corresponds to a desired peak with time index 2. Fig. 22 shows a block diagram for determining the parameter P for a peak filter . Fig. 23 shows a curve for the filter where P can vary with the peak position for a filter with a fixed duration.

Fig. 24 shows a curve for how the parameter

P for a peak filter can vary with the filter length (or memory duration) for a fixed peak position.

Fig. 25 shows curves for how the parameter

P for a peak filter can vary with the filter length and the filter's time delay.

Fig. 26A schematically shows a pulse time code system

in accordance with prior art.

Fig. 26B schematically shows a pulse time code system for the present invention where the size of the parameter that is transmitted is represented by the time interval between successive short single pulses of essentially constant duration.

Fig. 26C further schematically shows the pulse time code system of fig. 26B. Fig. 26D schematically shows a pulse time code system of the type shown in fig. 26B and 26C but where three-pulse groups are used. Fig. 27 shows a block diagram of a code converter which enables a system to receive signals of pulse-time code form in three-pulse group.

Fig. 28A shows a block diagram of selector 316

on fig. 27 in more detail.

Figs. 28B, 28C, 28D and 28E show curves to explain the operation of the circuit of Figs. 28A. Fig. 29 shows a block diagram of circuits for producing the three-pulse groups of fig. 26D. Fig. 30 shows a block diagram for generating the pulse time code according to the invention.

In fig. 3 shows a normal oil drilling rig 20 with a rotary drill 21, a kelly 22, a hose 23, a stand pipe 24, a drill stem 25 and a drill collar 26. A mud pump 27

and a mud reservoir 28 is connected to the rig in the usual way to deliver drilling mud under pressure to the standpipe. Mud under high pressure is pumped down through the drill stem 25 and the drill collar 26 and through a special remote measuring device 50

and to the drill bit 31. The drill bit 31 is provided with a normal drill jet device 33. The diameter of the drill collar and the remote measuring equipment 50 is shown large and not in proportion to the drill stem 25 for the sake of clarity. The drilling mud is circulated down through the drill stem as shown by arrows and up through the annular space between the drill stem and the borehole wall. When the mud again reaches the soil surface, it is brought back to the reservoir where drilling residues sink to the bottom and the mud is filtered before being recycled to the pump.

Between the drill bit 33 and the drill collar 26, the remote measuring equipment 50 is embedded. The remote measurement equipment 50 comprises a housing 250 which contains a valve unit or simply a valve 40, an electronic processing unit 96 and sensors 101. The valve 40 is arranged for the momentary passage of mud from the inside of the drill stem to the annular space 60. Normally (when the valve 40 is closed) all the drilling fluid must be driven through the jets 33 and exerts a significant mud pressure on the stem pipe 24 in the order of 140 to 210 kg/cm 2. When the valve 40 is open as a result of a signal from a sensor

101 and the electronic processing unit 96, some sludge is passed by and the overall resistance to the sludge flow decreases temporarily and a pressure change can be detected in the standpipe 24. The electronic processing unit 96 produces a coded sequence of electrical pulses representing the parameter measured by the sensor 101 and the corresponding opening and closing of the valve 40 are produced with the result that the corresponding pressure pulses occur in the standpipe 24.

A pressure converter 51 is arranged on the ground surface which produces electrical voltage which represents the pressure changes in the stand pipe 24. Signals which represent these pressure changes are processed in an electronic unit 5 3 which produces signals which are suitable for recording in a recording device 54 or another recording device. The recording device 54 is operated in accordance with the depth of the drill bit in the borehole.

Fig. 4A shows in detail the telemetry equipment 50

on a natural scale. The housing 250 contains the valve 40, the electronic processing unit 96 and sensors 101, and is divided into two parts 250a and 250b. The upper part 250a above the dashed line 249 contains the valve unit 40 with the associated mechanism and will be explained in more detail below. The lower part 250b below the dashed line 249 contains the electronic processing unit 96, the sensors 101 and associated mechanisms and will be explained in more detail below. The drilling mud circulates downward past the housing 250 as shown by arrows 65, through the jet nozzle 33 and back as shown by arrows 66 to the ground surface through the annular space 60 to the reservoir 28. The valve assembly 40 om-

holds the valve body 68 and the valve seat 69. The valve body and the valve seat are arranged so that the cross-sectional area A of the valve body is somewhat larger than the cross-sectional area B of a compensating piston 70. Therefore, when the pressure in the chamber 77 is greater than in the chamber 78, the valve body 68 will be forced downwards and the valve 40 will therefore close itself more tightly with increasing differential pressure.

The mud pressure in the chamber 77 is at all times substantially equal to the pressure inside the drill collar 26 in fig. 3 and 50

on fig. 4A, due to the opening 77a in the wall of the housing 250. A mud filter 77b is arranged in the opening 77a to prevent solid particles and drilling debris from entering the chamber 77. When the valve 40 is closed, the mud pressure in the chamber 78 is equal to the mud pressure in the annular space 60. When the valve 40 is open and the pump is working, the sludge will flow from the chamber 77 to the chamber 78 and through an opening 81 to the annular space 60 with a corresponding pressure drop.

A double-acting electromagnet 79 is arranged to open and close the valve 40 in accordance with electric current supplied via lines 90.

It is assumed that P^q indicates the mud pressure in the annular space 60, P^7 in the chamber 77 and P?g in the chamber 78. When the valve 40 is closed, P7g = PgQ. When the pump 27 is working and the valve 40 is closed or nearly closed, V-j-j will be greater than P^g and the valve body 68 is forced against the valve seat 69. When the valve 40 is open, i.e. lifted in the drawing, sludge will flow from the chamber 77 to the annular space 60, and because the resistance to the current in the opening C (Fig. 4B), <P>^7 = P?g > PgQ. The frames 83 and 94 are filled with oil of very low viscosity, 5 centistokes or less and connected to the passage 86. The floating piston 82 causes a pressure Pg3 in the oil-filled chamber 83 and this pressure is at any time equal to Pg. Therefore, at all times P7g = Pg3 = Pg4- When the valve 40 is open and because P^g = <Pg>^ and P77<>> Pg^, the valve 40 will be forced into the open position by a force F = (area B) (P77 - P34)• The valve 40 can therefore be described as bistable, i.e.

when it is open it will tend to stay open and when it is closed it will tend to stay closed. When the valve is almost open, it will tend to move to the open state and when it is almost closed, it will tend to close. The valve 40 can therefore be swung from one position to another using relatively little energy. The valve function can be seen as a mechanical equivalent to a bistable flip-flop circuit.

Fig. 4B shows the valve 40 in the open position, while Fig. 4A shows the valve in the closed position.

In fig. 4A, an electric pressure switch 91 is arranged which is closed when P77 > <P>7<g> (when the pump is working) and broken when P77 = P7g (when the pump is not working). A line 92 leads from the switch 91 to a power source 93 and can therefore switch the power source on and off. By means of an electronic counter 94 and an electromagnetic turner 95

each of the four sensors 101 can be connected in turn to the electronic processing unit 96 and in turn stop and start the sludge pump 27 or stop the pump or pumps in accordance with a predetermined code that can be predetermined by means of an element 94 .

In fig. 5A shows a sensor 101 which e.g. can sense gamma rays, temperature, pressure, gas content, compass direction, inclination, magnetic field or can be a gyro-compass, which can be connected in turn as mentioned above. A gamma ray sensor can be designed as an ionisation chamber, Geiger counter or scintillation counter. All of these can be arranged to supply a DC voltage proportional to the gamma ray flux hitting the sensor.

It is clear that the switching from one type of sensor to another is carried out by the inverter 95 which is electronic rather than mechanical.

The sensor 101 is connected to an analog-digital converter 102, which is connected to a processor 103 which is in turn connected to a power source 104. The power source 104 is connected to the windings 105 and 106 in the double-acting electromagnet 79 in fig. 4A. It all works as follows: The sensor 101 produces an analogue electrical signal represented by the curve 101a which shows the sensor's output signal as a function of the depth of the remote measuring equipment 50 in the borehole. The converter 102 converts the analogue signal into digital form by successively measuring the size of a large number of ordinates in the curve 101a and converting each of these into a binary number represented by a binary word. The curve 101 can show variation of the signal from the converter

in hours, while the curve 102a only represents a single ordinate. The time scale on the abscissa in fig. 102a will be in seconds and the entire curve 102a represents a binary 12 bit word and is actually the decimal number 264 9. Each 12 bit word on curve 102a represents a single ordinate on curve 101a. The usual binary coding includes time gaps between each binary word. After the pause, a start pulse is sent to indicate the beginning of a new time interval for a binary word. This start pulse is not part of the binary word, but serves to indicate that a binary word is beginning. The binary word which is then transferred and which is an indication of the value of an ordinate in the curve 101a, is followed by a pause which is again followed by the next binary word which represents the size of the next ordinate etc. in quick succession. The continuous curve 101a thus represents a series of binary numbers or words, each of which represents a single point on the curve 101a. It is important that there is always a pause between each binary word where no signal is transmitted and often several binary words are long and are used for an important purpose which will be explained in more detail below. For decoding on the earth's surface, a time generator 1 is arranged which must be very constant and synchronous with a time generator 212 or 309 on the earth's surface to produce a series of pulses at regular intervals.

The curve 103a represents a single bit in the binary word 102a and the abscissa is here again very different from the previous curves. The time axis of curve 103a is in milliseconds, because the curve only represents a single bit. Each individual bit is transformed into two electrical pulses with a duration t and are separated by a time interval t.

x y The curve 104a is a repetition of the curve 103a and is greatly amplified with the help of the power source 104. An electrical pulse 104b is supplied to the winding 105 for opening the valve and electrical pulses 104c are supplied to the winding 106 for closing the valve. The valve 40 in fig. 4A is opened by pulse 104b and closed by pulse 104c and remains open for approximately time t. The time tx is set so that it provides the correct energization of the windings and the time t is set to open the valve 40 for the correct period of time. Both of these times are determined by a time generator 2.

The telemetry information from a sensor to the earth's surface occurs with suitable pauses between the transmissions of the binary words. As a result of these pauses, it is possible to store in a storage device on the earth's surface noise that is due to the drilling operation alone without a valve wave. This will be explained in more detail below.

As already mentioned above, the valve 4 must 0

on fig. 4A be very fast-acting and this requires considerable force. Experiments have shown that 1/2 to 3/4 horsepower is needed to achieve the required speed.

This power is only exercised for very short periods of time

and with a power of 1/2 horsepower for approx. 40 milliseconds will give it sufficient energy for a single operation of the valve. This energy can be around 15 Joules.

A battery that is sufficiently small to be placed in the housing part 250b can deliver approx. 4 million Joules without recharging or replacement. The system is therefore able to perform 130,000 complete valve operations (opening plus closing). In reality, the energy consumption is less than 15 Joules per operation. The self-induction, Q, and the movement impedance of the windings cause a current build-up which is relatively slow as shown by the curve 272A in fig. 5C and 300, 301 in fig. 6E. The total energy per pulse is significantly less than .15 Joules and has been measured down to 9 Joules so that the capacity is 216,000 complete valve operations. A larger capacity can be obtained by using the circuits described below with reference to fig. 5C. From the above, it appears that the necessary energy from the batteries down in the borehole makes practical remote measurement very applicable. Achieving the necessary very large energy (1/2 horse power) represents difficulties, however.

It is clear that the solution of such a problem

must involve the storage of energy in a mechanism which can trigger this suddenly within a very short time and thereby provide the necessary short bursts of great power.

Such a mechanism must have a hammer effect and has previously been achieved by using compressed air, pre-tensioned springs 0. 1. Storing energy in a capacitor requires a capacitor with a large capacity because the energy that is stored is primarily dependent on the capacity and the square of the stored voltage, and since small self-induction, fast action and energizing windings are necessary, the necessity of low voltage becomes apparent, which means that very large capacitors will be necessary.

Mathematical analysis and experiments have led to the following optimal circuit parameters: 1. The self-induction in the windings 0.1 henry in the energized state and 0.07 henry in the non-energized state, i.e.

the windings are made with outlets.

2. The resistance in the winding 4.5 ohms.

3. The voltage at which energy is stored is 50 volts.

4. Condenser capacity 10,000 mfd.

5. Power supply to the drive circuit 10 amp.

In order to achieve a quick energizing effect, windings with low self-induction are desirable. It has also been determined that the power supply to the drive circuits can be increased beyond 10 amps. However, low voltage requires a very large capacity.

Recent advances in so-called molten salt batteries have provided energy sources that are very compact and capacitors with very high capacity, namely 10 mfd with a volume as small as 16 cm 3. This is unacceptable because it requires heating to 500°C which is impractical and the costs are far too great. But further progress has led to a tantalum capacitor which is well suited to the purpose.

From the above it appears that at least 216,000 complete valve operations can be achieved from one battery charge. If it is assumed that the telemetry system can deliver continuous data by sending five pulses per minute, the system can be in operation in the borehole for 440 hours. However, it should be pointed out that continuous operation is not necessary. The measuring equipment can be used by simply sending an order under the control of the switch 91 and the elements 94 and 95 in fig. 4A.

Furthermore, as will be explained in more detail below, the circuit shown in fig. 5C, an even greater number of valve operations is achieved with a pulse rate of 1 pulse per Sec. which seems practical.

There is also another parameter that must be determined, namely charging of the capacitor after discharging. The capacitor can be charged through a resistor connected to the battery, but this has sometimes been shown to be slow because the capacitor is partially charged and the current through the resistor decreases, and at the end of a charging period the charging current is almost zero. If the resistance value of the resistor is very small, the charging current will be too large, so that it will exceed the life of the battery. The best way to charge the capacitor is with constant current. The capacitor will then be charged with a maximum current corresponding to the maximum discharge current for the battery type used, in order to achieve maximum charge. With the right choice of charging current, a significant increase (by a factor of 2 or 3) can

of the energy available from the selected battery type. Constant current energy sources are well known and can be easily achieved with electronic integrated circuits within

a large current area.

Fig. 5B shows a power source that can be used in the drive unit 104 in fig. 4A to supply the necessary energy to the windings 79. In fig. 5B indicates 450 a battery or a turbo generator or other source of direct current, 451 a device for constant current and 452 the capacitor. The capacitor is charged from the device 451 and discharged via line 453. Line 454 supplies the regulated current for the electronic equipment down in the borehole.

A device is also provided in the event that the valve is left in the open position for a longer period of time due to a malfunction. This device for automatically closing the valve in the event of a malfunction is indicated by 269 in fig. 4A and shown in detail in FIG. 6A, 6B and 6C.

It has previously been mentioned that the valve has a hydraulically bistable action, ie when opened by means of winding 105 it tends to remain open and when closed by means of winding 106 it tends to remain closed. It is then possible as a result of an electrical or mechanical malfunction that the valve remains open. It should be noted that if such a malfunction occurs, the drilling operation may continue even though some wear may occur in the opening 81 of FIG. 4A. However, a disturbance of the hydraulics of the mud system when the valve is open for a long time is not desirable, and even if the drilling can continue, it is advantageous that the valve is closed most of the time and only open to produce short pulses necessary to produce hydraulic shock waves.

In fig. 6a, 6b and 6c, the rod 100 is used for

to push the valve into the closed position by exerting the necessary force downwards on the rod 80 in fig. 4B.

The upper end of the mechanism is exposed to the mud in the drill stem, i.e. is under hydraulic pressure plus the pressure difference across the drill bit, i.e. the pressure difference between the inside of the measuring equipment 50 and the annular space 60. When the pump is not working, the pressure in the zone 111 is only hydrostatic and when the pump works, the hydrostatic pressure plus the differential pressure occurs. As the differential pressure is in the order of magnitude of 70 to 140 kg/cm 2 , a large pressure change occurs in zone 111 when the pump is started, i.e. a pressure increase of 70 to 140 kg/cm 2 . When the pump is not working, the zones 112,113 in fig. 6A the same as the pressure in the annular space because the pipe 114 is connected to the chamber 84 containing oil with the pressure in the annular space (see Fig. 4A) and because the opening 115 is in communication with the zones 112 and 113.

If the pump is started, the pressure in zone 111 will increase 2

substantially, ie by 70 to 140 kg/cm and the piston 116 is pushed down against the pressure of the spring 107 (not shown in Fig. 6B) and the high pressure oil in the zone 112 pushes the piston 108 down and compresses the spring 110. When the pump starts, the situation will change from that shown in fig. 6A to 6B and both pistons 116 and 108 in the lower position and rod 100 projecting downward as shown.

As a result of the opening 115 and the action of the spring 110, the piston 68 is pushed upwards at a speed determined by the size of the opening 115, the spring constant of the spring 110 and the viscosity of the oil in the zones 112 and 113. This speed can be easily controlled to a desired value, e.g. so that the piston 108 will return to its original upper position within 1 minute. After 1 minute, the device will therefore assume the state shown in fig. 6C. For the same reason, when the pump is stopped, the action of the spring 107 and the opening 115 will cause the piston 116 to be lifted back to the original state in fig. 6A.

It is therefore clear that each time the mud pump is started, the rod 100 will be moved down a distance d as shown in fig. 6B and then return to the normal retracted state. As the pump is stopped in normal drilling every time a drill stem section is to be inserted, the rod 100 will make a simple downward movement and then return to its original upper position.

It has previously been mentioned that the rod 100 is arranged so that when it is pushed downwards, it will push the magnetic armature rod 80 in fig. 4A downwards and close the valve. The device in fig. 6A, 6B, 6C and 6D are thus a safety device, i.e. if the valve is left in the open position as a result of an electrical or mechanical malfunction, the valve will be forced to the closed state after a maximum of 10:1 meters of drilling.

Fig. 6D shows in more detail the device 269 in fig. 4A.

The hydraulic automatic closing system described in connection with fig. 6A, 6B, 6C and 6D will automatically close the valve every time the mud pump stops and is restarted and thus any mechanical locking of the valve can be remedied. There is one case, however, which requires further vigilance, namely if the electrical closing circuit 103,109 in fig. 5A fails for some reason, e.g. a burnt magnet winding, the valve can be opened again electrically shortly after the hydraulic automatic closing device has closed the valve.

Fig. 6E shows an electrical system which will

prevent operation of the valve in the event of an electrical fault with the downhole equipment. S. denotes the closing winding and S2 denotes the opening winding. The resistor 1^ is connected in series with a part of the drive circuit 104 which causes the closing winding S^ to be energized. The resistor R2 is connected in series with a part of the drive circuit 104 which energizes the opening winding S2> These resistors have a very small value (0.05 to 0.2 ohms). The operation of the system described in fig. 5A is as follows: The electrical opening current pulse is produced first and is shown in fig. 6E in the form of the pulse 300. The electrical opening current pulse is produced later, after a time t yog is shown by

301 in fig. 6E. It should be noted that the electrical pulses 300 and 301 represent current absorbed by the windings and not the supplied voltage, and the resistors R, and R2 produce voltage losses R-^i-^ resp. R2i2°g ^ i'^ 2 is the current through the respective windings, and consequently if one of the windings is burnt or the connection broken, no

current will flow and no pulse will be produced, as with any other electrical fault, no current will flow through one or both of the resistors and R,,.

The length of time t in fig. 6E and the length of tx is explained above in connection with fig. 5A.

The delay in the delay element 302 is equal to t. In other words, the delay element 302 delivers on

its output a pulse that is identical to the input pulse, but delayed by a time t . Such a delay system is well known and will not be described in more detail here.

Since the delay in the element 302 is equal to t , will

y

the pulse 303 occurs at the same time as the pulse 301.

An anti-coincidence circuit 304 (an OR gate circuit) delivers at the output 305 an electrical signal only when one of the pulses 301,303 is applied, but delivers no pulse when both pulses 301 and 303 occur. A relay 306, which is energized by the signal on the output 3 05, disconnects the energy to the equipment down in the borehole. Thus, if only a closing pulse occurs without an opening pulse or if only an opening pulse occurs without a closing pulse, the energy to the downhole equipment will be disconnected and the valve will be closed mechanically by the automatic hydraulic closing system described above in connection with Fig. 6D.

As an alternative to fig. 6E, the relay 306, which may be an electronic switch in the form of transistors, may be arranged to de-energize the circuits for the opening winding. This can have certain advantages because the closed circuits will continue to function, and one of the purposes is to

ensure closing of the valve. Furthermore, the electronic counter 314 can be placed between the OR gate circuit and the relay 306, so that an electrical malfunction will not disconnect the energy. The energy will then only be broken after e.g. 2,

4 or 8 consecutive malfunctions.

It has been pointed out above that a very quick operation of the valve 40 in fig. 4A is important. The necessary shock wave will not be produced if the valve operation is slow. As the valve and its drive mechanism have considerable mass, considerable force is required to open and close the valve within the time period considered desirable. This power on the order of 1/2 to 3/4 horsepower can be provided by a power source described above. As in all such designs, a certain amount of force is required to ensure that the valve always opens and closes as desired. The various electronic logic circuits and drive circuits shown in FIG. 5A is designed to deliver square voltage pulses 104b and 104c which have a duration of approx. 40 to 50 milliseconds to ensure that windings 105 and 106 are energized for sufficient time to ensure operation of the valve. Fig. 5E shows the voltage pulse 104b of fig. 5A in more detail. At time 0, voltage is suddenly applied to the drive circuit 104 and increases almost instantaneously to a value of 270 and maintains this voltage for 50 milliseconds and then drops again almost instantaneously to the value 0.

The movement of the valve can be done by measuring the current flowing through the winding as shown in fig. 5D which is an oscillogram that shows the current as a function of time, and can e.g. is measured as the voltage loss across the resistor R1 or R2 in fig. 6E.

It is the current through the winding that determines the force and the valve body 6 8 in fig. 4A because the ampere winding number determines the electromagnetic force. As the windings have self-induction, the current will not build up immediately when the voltage as shown in fig. 5E is pressed. If the winding were a simple self-induction, the current would build UP as shown by the dashed line curve 271. In reality, something else occurs, because there is a sudden movement of the armature in the magnet 79 in fig. 4B when the valve is opened or closed in the form of a reactive electromotive force due to the speed of the armature which rapidly changes (increases) the self-induction in the relevant winding. In fig. 5D shows the curve 271 when the armature and valve body 68 are blocked in the open or closed state. The: solid curve 271 in fig. 5D shows the flow as it really builds up when the valve is not blocked, i.e. when opening and closing correctly. The curve 272 for opening or closing is very similar to this. From the curve 272 it appears that after applying the voltage, the current gradually increases until it e.g. when a value of 4 amp. at time Tg after 20 milliseconds. Then a sudden drop to a low of 2.2 amps occurs. at time T. after 25 milliseconds. After T-^ = 25 milliseconds, the current increases again until it reaches the usual exponential curve to a value of about 10 amp. after approx. 6 0 milliseconds depending on the resistance in the winding such as e.g. may be 4.7 ohms.

From the curve 272 it is clear that the valve 40 starts opening and closing at time Tg = 20 milliseconds and ends the movement at time T-^ = 25 milliseconds. As mentioned above, the opening and closing of the valve will begin at time Tg = 20 milliseconds and be finished at time T, = 25 milliseconds.

It is important that the time = 25 milliseconds depends on several factors. Thus, at high differential pressure T, will be greater than 25 milliseconds, e.g. 30, 35 or 40 milliseconds. Suffice it to say that the time T, indicates the time when the valve operation has ended and the current between T, and 50 milliseconds is actually wasted because the operation of the valve has already ended. This extra time is a safety factor to ensure that even under difficult conditions the valve will always be operated when the voltage pulse is applied.

In practice, the signal is used at time T-^

for disconnection of any other current to the magnet 79. Accordingly, all current between time T. and 50 milliseconds will be saved and thus reduce the amount of energy required to operate the valve 40. It should be noted that full safety factor is maintained and the current will continue to be supplied until the valve has finished its movement.

The electronic circuits used to carry out the above-mentioned are shown in fig. 5C where 104 is the drive circuit of fig. 4A. A resistance R^ or R^ with a small resistance value compared to the resistance in the windings, e.g. 0.2 ohms. The voltage across this resistance is therefore proportional to the current supplied to the relevant winding 104,106. Two circuits as shown in fig. 5C are required, namely one for the opening winding and one for the closing winding, but for simplicity only one of these circuits is shown. From the connection point between the drive circuit and the resistor, one ordinary amplifier 273 is arranged whose output voltage has the course of the curve 27 2a which is a reproduction of the curve 272 in fig. 5D. A derivator 274 produces an output signal that is proportional to the first time derivative of the input voltage, which derived voltage is shown by the curve 275. From the curve 272 or 272a it appears that the derived curve is always positive with the exception of the times Tq and T^ when it is negative. On the curve 275, the pulse 276 is negative. A conventional rectifier 277 only passes through the pulse 276 as shown by the curve 278. A delay circuit 279 produces an output pulse 276b which is a reproduction of the input pulse but delayed by the time T^-Tg. The pulse 276b as shown by the curve 280 thus occurs only slightly later than the time T^. This pulse 276b is supplied to an electronic switch 281 which interrupts the energy supply to the drive circuit 104 and stops the current immediately after the valve 40 has finished its operation. The switch 281 is designed to restore the power supply to the drive circuit 104 after a desired period of time. This is repeated when the next pulse 104a or 104b occurs.

It is important to note that energy can be saved in this way. As very large energy is required to operate the valve 40 at the necessary high speed, this energy saving is very important and it can be shown that the battery's lifetime can be increased as much as 5 times.

An important feature of the invention is its applicability, as it can easily be placed in an existing drill hole: 11. With previously known equipment, large special steel housings with a length of 10 meters or more and 20 cm in diameter are required to house it complicated instrumentation and the transport from place to place require specially designed vehicles. In the case of the invention where the valve mechanism is embedded in the main mud flow, it is possible to eliminate a heavy, very long and expensive special housing as only a short piece of the drill collar is necessary. In a practical embodiment of the invention, this part is only approx. 90 cm long and 17 cm in diameter.

One of the important features of the invention is therefore that no heavy and long house is necessary. This is particularly advantageous when magnetic measurements such as compass directions for controlling drilling in directional boreholes are to be made and where a non-magnetic drill collar is required. Non-magnetic drill bits are not only heavy, 2-3 tonnes, but also very expensive, up to NOK. NOK 100,000 because they must be made from an absolutely non-magnetic material such as K Monel. With a device according to the invention, a standard drill collar with an outer diameter of 15-23 cm can be used. As a result of simplicity, dimensions and coaxial construction of the valve system according to the invention and the associated parts, a special advantage must be achieved in that all the necessary drive energy and associated equipment can be placed in a pressure-proof pipe with a sufficiently small diameter that it can be accommodated inside a normal drill collar without unduly affecting the mud flow. Some sensors must be placed as close as possible to the drill bit. Gamma ray sensors in particular must be able to detect the penetration of the drill bit into a given soil formation as soon as this happens. Some sensors such as inclinometers require accurate indication in relation to the attack surface of the drill bit used for directional drilling, and must be placed at a significant distance from magnetic or paramagnetic materials and the housing 250a and 250b must lie exactly flush with the drill collar.

The drill collar is equipped with indication 253a and the angle for this marking must have a constant angular relationship to the marking 254a which is placed on the remote measuring equipment. This known angle is then introduced into the calculation to determine the direction and angle in relation to the vertical plane for the drill collar.

Fig. 7A shows the remote measurement equipment 50 with a housing

250 with an upper part 250a and a lower part 250b as previously shown in fig. 4A. The upper part 250a is arranged inside a short piece 254 which is approx. 90 cm long. This short piece has an internal bore with a diameter of e.g. 12 cm sufficient to house the valve unit 4 0

and also to allow an unrestricted flow of drilling mud past the upper portion 250a through a passage 61 also indicated by 61 in FIG. 4A. The housing part 250a has a small diameter, preferably approx. 7 cm or less. A drill collar 255 is usually approx. 10 meters long and has a lot of weight and is expensive. The inner diameter is approx. 8 cm + 1.5 cm. The central one

part 256 is provided with the lower housing part 250b and has a somewhat smaller diameter than the upper housing part. Small clearance is very important to prevent oscillation when the equipment vibrates during drilling. The outlet opening 85 is the same as shown in fig. 4A. The housing 250b is suspended inside the part 254 by aids which are not shown in the drawing. The cross-sectional shape of the centering portion 256 is shown in FIG. 7b with a groove or passage 258 for free flow of the drilling mud.

The lower housing part 250b consists of several parts which

are connected to each other by aids shown in fig. 7C, where each part is provided at the top with a recess 26 0 and at the bottom with a projecting part 261 which fits into the recess of the part situated below. The parts are held together by an internal connecting sleeve 262 which is attached to the parts by means of screws 265 through holes 263, 264. Such a connection ensures accurate angular position when the part 253 is bent.

As mentioned above, the angle between the marks must

253a and 254a be known to calculate the angle in relation to the vertical direction for the bent part. It is also necessary that the angular displacement between the axis of an inclined

indicator and the marking 254a are known and at regular intervals during drilling it is preferred, but not necessary, that the angle between one of the horizontal axes of the inclination indicator and the marking 254a is zero. For this purpose, the housing part 250b is equipped with angle-indicating projecting parts 261 as shown in fig. 7A and 7C.

A necessity in the present invention

is to achieve hydraulic shock waves that have an effect analogous to the hammer effect in water. By suddenly stopping the sludge flow, the pressure is suddenly increased. This local increase in pressure is propagated through the mud flow in the manner of a hammer effect. It is well known that a sudden and local change in velocity produces a corresponding local change in pressure, and consequently a sudden and local change in pressure will produce a sudden and local change in velocity. As a result of elasticity and inertia in the fluid, the changes are transferred from the part where they are originally formed to neighboring parts with a speed corresponding to the propagation speed of the compression waves. The propagation of shock waves is a very complicated problem. To satisfy practical requirements, a parameter must be determined that is most representative of the task of achieving a clearly defined shock wave. Two parameters K. and K2 must be considered in more detail. When one or the other of these parameters exceeds a suitable value, a clearly defined shock wave is produced.

The parameter K-^ is the average rate of change for the speed of the mud flow through the valve when it is open. If V(t) is the speed of the mud flow through the valve as a function of time in cm per second.

At time t = 0 when the valve starts to open, the velocity is 0, i.e. V(0) = 0. At time t = T a. ^ when the valve is fully open and the velocity of the valve actuation is Vf which is the passing velocity during full flow , is

and consequently the parameter K, which is the average rate of change, is the rate during the time period T a^:

2

measured in cm/sec.

If it is assumed that the parameter K. exceeds a certain threshold value, i.e. when: it is determined that:

The parameter K2 represents the mean velocity change of the area of the opening in the valve in T(<v>)

the time interval T .

a

In the previously stated equation (1), S(t) was defined as the area of the valve that is open at time t, so that t = 0 gives S(0) = 0 and t = T cL so that:

where Sg is the total opening of the valve. The parameter

K 2 is then:

It is assumed that the parameter K2 exceeds a suitable threshold value, i.e. when:

and a clearly defined shock wave is formed. Attempts have

2

shown M = 100 cm /sec.

Roughly speaking, K is proportional to K,,. The parameter K2 is more applicable because it directly indicates the construction and operation of the valve.

There is also a parameter (see B^C^ on fig. 2A) that must be taken into account when evaluating the curves on fig. 8A to 8E. Each of these curves corresponds to a set of numerical values of K, and or K2 and T, .

Figures 8A to 8E show the effect of varying

Kl and e^--'-er K2Qg Tb^<V>^ ^or ^ cause transfer from an area of small pressure variation to an area of hydraulic shock waves. Each of these curves shows how the pressure detected on the earth's surface as ordinate varies with time t as abscissa.

9

The area of the opening was 0.5 cm" in these experiments. Experimental data were obtained at several boreholes drilled to conduct remote sensing experiments.

When carrying out the above-mentioned experiments, a large selection of existing mud pumps and various interfering effects were taken into account, and pump pressure variations for a given mean mud pressure varied a great deal from experiment to experiment. Eliminating large interfering mud pressure signals was complicated. The pump pressure signals could differ by 10 to 20 times. In order to determine optimal values for K2 or and T^ ^ the drilling was stopped and a very good pump was used. Figs. 8A to 8E are therefore not representative of a typical condition, but represent a condition where different noise (from pumps and other sources) was reduced and then an average value was taken by calculation to obtain maximum values for the parameters K2 or K^ and T^^.

The corresponding values for these parameters are for fig. 8A to 8E given in the table below:

Figs. 8A to 8E represent mean values from a wide range of samples. In these tests, a standpipe pressure of 210 kg/cm 2 was used and the variations in the pressure were within the range of 7 kg/cm 2. These tests were carried out with different types of valves such as motor-driven valves, rotary valves, poppet valves, etc. Fig. 8F shows exactly the standpipe pressure recorded in a test at a depth of 3000 meters and with a standpipe pressure of 200kg/cm 2. Fig. 8A shows a slow-acting valve. The numerical values for the parameters in fig. 8A is K2 =

0.5 cm<2>/sec. and Tfa^ = 2 sec., i.e. they correspond to what is assumed by the previously known technique shown in fig. IA and IB. Accordingly, both Fig. 8A and IB the case of slow pressure pulsations. On the other hand, fig. 8E the application of a quick-acting valve where K2 = 100 cm<2>/sec. and Tb<*v*> = IO-<1> sec. Accordingly, fig. 8E hydraulic shock waves and the valve wave of fig. 8E is very similar to the valve wave of FIG. 2B.

Figures 8B, 8C and 8D show different transitions from slow pressure variations to hydraulic shock waves.

In the samples shown in fig. 8B, 8C and 8D are as similar as possible. The measuring equipment was located near the bottom of the borehole at approx. 2,400 meters and the mud viscosity was approx. 40 Funnel with a weight of 1.3 kg/litre. The open valve had a cross-sectional area of 0.7 cm 2 . The normal stand-pipe pressure was 210 kg/cm 2 and the valve used was similar to valve 40 with a modified and slower operation without bistable action, i.e. the valve was a simple pressure balanced valve and the flow rate was controlled by narrowing the inlet passage. It should be noted that the valve operation in fig. 8B was relatively fast, but did not produce the desired hydraulic pressure waves. However, the rapid opening indicates that rapid operation is desirable. The outlet velocity was in the order of 19 litres/sec.

Adjustment of the inlet area and the outlet area

and the electrical energy supplied to the windings, different valve operating speeds can be achieved. From the above it appears that no shock waves are produced when Kbr9 i= nge0s ,5 nåcm r 2/se= k.1, 0o0 g cm ti/lnsæekrm.et ideal shock waves are produced

Another parameter expressing the necessity of the strength of the shock wave shall be introduced below. Two different conditions must be taken into account. One parameter K. applies to the amount of sludge in cm 3 that passes the valve in the time period Ta ^ • The other parameter represents the average flux in the sludge flow in the time period T (<v>), so that

To simplify the problem, it is assumed that the speed increase for the current in the time period Tg^ is constant and equal to K^, so that:

It is also assumed that the increase of the valve opening is constant and equal to K_, so that:

Consequently, the volume of the sludge that passes the valve in the time period T ^ will be:

3.

The parameter K-, is thus the amount in cm 3 of sludge that passes the valve in the time interval Ta^<V>^• This is the volume of interest for the period of a single opening and closing of the valve. Another alternative is, instead of the parameter K-, to use the parameter K. which represents the flux of the sludge flow in the time period T cl ^, i.e.:

Decoding of the signals from the pressure converter 51 will be considered below. Fig. 9 shows the apparatus and fig. 10A to 10G show waveforms when decoding the signals.

The signal from the pressure converter 51 comprises an information-carrying signal together with interfering signals which obscure the usable signal. The information-carrying signal represents a coded message which is obtained by means of the valve 40 in accordance with a sensor. One of the interfering signals is produced by the pump 27 and contains a strong component for the mud pressure produced by the pump. This component is due to the circulation of the mud through the drill stem and back into the annular space between the drill stem and the borehole wall. Superimposed on this is an alternating component produced by repeated movement of the pump's reciprocating pistons.

In order to improve the reception, it is desirable to remove from the output signal from the converter 51 and to remove the continuous pressure component produced by the pump 27. For this reason, a frequency selective filter 150 is connected to the converter 51 for passing frequencies in the range 0.1 to 10 Hz and attenuate frequencies outside this range. Frequencies that apply to the constant pressure component are below 0.1 Hz.

In the description, a distinction must be made between the term filter as used for the frequency-selective filter 150 and a digital filter, which will be described in more detail below. With a filter such as the filter 150, normal filtering is carried out using analogue electronic networks. The term filter can be used for a wave filter, such as a Shea filter, see e.g. "Transmission Networks and Wave Filters" by T. E. Shea, D. VanNostrand Co., New York, N.Y. 1929. On the other hand, a digital filter can be an adaptive filter, a pulse-shaping filter, a peak-shaping filter. The output signal from a digital filter is obtained by processing the digital input signal with the filter's weighting coefficients. A digital filter is a computer.

Signals delivered to output terminal 151 from filter 150 are the following function of time:

where B(t) is the information-carrying signal, P(t) is the interfering signal due to periodic pressure variations from the pump, and U(t) represents ambient noise. The ambient noise is produced by various effects such as the teeth on the drill bit during drilling of gears in connection with the drill stem and by other devices during the drilling operation. In some cases, U(t) is an approximation to white noise, but in other cases can differ significantly from this.

The coded message expressed by the information-carrying signal B(t) is a series of binary words and each of these contains a series of bits. A single bit in a binary word is produced by a single actuation of the valve, i.e. a single opening and closing. Such a simple operation produces a hydraulic shock wave which on the ground surface produces a single valve wave as shown in fig. 2B. The message expressed by B(t) therefore takes the form of a coded sequence of valve waves. Figures 10A to 10G show various steps that must be performed to separate the information-carrying signal B(t) from the interfering signals. To facilitate the explanation, B(t) in fig. 10A expressed by a single valve wave or a coded series of valve waves. The valve wave in fig. 10A is therefore of the same type as the valve wave in fig. 2B. However, there is a slight change, as the marking with "s" on fig. 2 is the loop in fig. 10A. In fig. 10A to 10G, the different times are denoted by t-^,<t>2» <t>15't16 without any marking with "s". For the sake of clarity and to facilitate explanation, the time scales for these curves are distorted.

In order to eliminate interfering noise signals and produce a signal that represents the coded message, the following operational steps are carried out one after the other: Step 1: In this step, a signal with three components shown in fig. 10A transformed into a two-component signal as shown in FIG. 10C. The purpose of this step is to eliminate the pump noise P(t). As a result, the valve wave shown in fig. 10A transformed into a double valve wave as shown in fig. 10C.

Step 2: The purpose of this step is to eliminate the ambient noise component.

Step 3: In this step, each double valve wave as shown in fig. 10D converted to a single signal pulse as shown in FIG. 10G. Consequently, a coded sequence of single pulses is obtained which represents in digital form the parameter measured by the sensor 101 at the desired depth in the borehole.

Fig. 10A shows three components of the signal F(t) as expressed by equation (22). These components are the valve wave B(t), the pump noise P(t) and ambient noise U(t). As previously mentioned, the signal F(t) is obtained by means of filters 150. The filter is connected to a delay element 152 which delays the input signal F(t) by a time Tp which is a period of the oscillation produced by the pump 27. The signal on the output conductor 153 from the delay element 152 can be expressed as F(t-T p). The three components of this signal are shown in fig. 10B and is the delayed valve wave B(t-T P ), the delayed pump noise P(t-T ) and the delayed ambient noise U(t-T ). time-

P. P

the space T Ir depends on the pump's rotation speed and as this is not constant, the delay T Pen is a variable delay. A suitable control of the delay element 152 must therefore be provided in accordance with the rotation speed of the pump 27. Therefore, the delay element 152 receives via the conductor 154 a timing pulse from a pulse generator 155 which is driven mechanically by the pump and produces a suitable number of pulses for each revolution of the pump. A drive chain 156 is provided for this purpose.

It is assumed that the pump 27 makes N-^ strokes per second,

and therefore T = 1/N^. The pulse generator 155 delivers time control pulses with a relatively large repetition rate N2 which is a multiple of N^, and therefore N2 = KN^, where

K is a constant that is chosen equal to 512. If, therefore, the pump strokes are one per second, the pulse generator must produce 512 pulses per second. It is clear that the pulse rate of the mud pump 27 varies with time and therefore N2 will vary to ensure that the delay caused by the delay element 152 is always equal to one period of the mud pressure oscillations produced by the mud pump 27.

The signal F(t-T ) from the delay element 152 is supplied to the input line 153 to a subtractor 160 which also receives on the input line 161 the signal F(t) from the filter 150 and produces on the output line 162 a difference signal

Since P(t) is periodic with a period T , becomes

As a result of the periodicity of the pulsations produced by the pump 27, noise caused by the pump is eliminated and the signal on the output conductor 162 from the subtractor 160 can be expressed: for the information-carrying signal and

for the ambient noise signal.

Both the information-carrying signal b(t) and the noise signal u(t) are shown in fig. 10C. By carrying out step 1 as explained above, the information-carrying signal B(t) as shown in fig. 10A transformed from a valve wave into a different information-carrying signal b(t) as shown in fig. 10C. The signal b(t) can be described as a double valve wave in contrast to the signal B(t) which represents a single valve wave. A double venti wave comprises two half venti waves designated "A" and "B" in fig. 10C. These valve waves are separated from each other by the time interval T P. The valve wave "A" is similar to that of Fig. 10A, while the valve wave "B" represents the inverted form of the valve wave "A".

The signal x(t) in equation (25) is further supplied to the input conductor 162 in an analog-to-digital converter 163 which

is controlled by a time control generator 178. On the output conductor 164 from the converter, the signal is obtained:

where in accordance with the symbols used here x^_, b^_ and

u^ are digital versions of the analog signals

<>> x(t), b(t) and u(t). The signals x^ and ufc have the form of time series:

and the signal bfc is a valve wave of finite length: The signal containing the double valve wave bfc and the noise signal ufc is applied to a digital filter 170 of length (n + 1) with a memory function

A digital filter is used here in the form of an adaptation filter and a memory function afc is used to optimize the effect of the filter. Optimal conditions are achieved when the signal-to-noise ratio at the output of the filter 170 has its maximum value.

The memory function a^_ of the filter 170 is controllable to ensure at any time that optimal conditions are achieved during measurement. The control of the filter takes place by means of a computer 172 which receives suitable data from a storage device and an outlet element 173 as will be described in more detail below.

A signal y appears on the output conductor 174

from the filter 170 and can be expressed as consisting of the input function x and the memory function a^, so that

where the star denotes an enclosure. By setting xfc = b^ + ufc in equation (33), becomes: which is the filter response to a clean input signal and:

is the noise signal. This is shown in the functional diagram in fig. 11.

To ensure optimal conditions of the filter 170, a specific time is selected, e.g. t = tQ and it is necessary that the instantaneous value of the signal appearing at the output of the filter at the time t = tg is as large as possible in relation to the mean value of the filter noise at this time. To detect the signal cfc in the filtered output signal ufc, the signal-to-noise ratio must be:

If the signal (bg,b^,...,bn) of length (n + 1) is processed in a filter of length (n + 1), the series (Cg,C^,... c , ...c„ is obtained i,c„ ) with length (2n + 1) where c is the

n 2n-l 2n 3 n

central value in the series. At time tQ = t ,p becomes

where E {v<2>} is the mean value for the noise.

It is assumed here that the ambient noise u is white noise and that the maximum value of the signal-to-noise ratio u can be achieved when:

where k is chosen equal to one. The presence of a signal in the noise when it comes to white noise, optimal conditions can be achieved when the memory in the filter is given by the inverted signal, namely with the coefficient feature (bn'bn-l""V*

The memory in the filter 170 is determined at all times by means of the computer 172 which is connected to the filter via a channel 175. The term channel denotes here suitable conductors, connections or transmission means to the extent necessary. The element 173 serves to store the function b for subsequent transmission via the channel 176 to the computer 172. The computer's function is to reverse input data expressed by the row (bg,b^,..-b ) in order to output the row (bn,b^ ,...bg) which is then supplied to the adaptation filter via channel 175 and is stored there as memory for the filter in accordance with equation (39).

The filtered output signal y appearing on the output conductor 174 from the matching filter 170 is supplied to a digital-to-analog converter 181. Since the signal y is a signal of digital form, the analog function on the output conductor 18 2 from the converter will be an expression in accordance with the symbols used as y(t).

It should be noted that making the signal-to-noise ratio as large as possible in the filtered output signal y^_ is equivalent to making the noise signal v as small as possible compared to the information-carrying signal c so that:

The output function y(t) from the adaptation filter~ as shown in fig. 10D is very close to the function b(t)

which is shown in fig. 10C.

An important feature of the invention involves the storage of the function bfc by means of the element 173. What is necessary for the storage of b^ will be explained in more detail below with reference to fig. 12 and consists of the following six steps.

Step a. The drilling is stopped, i.e. the drill bit 31 is lifted a short distance from the bottom of the borehole, the drive is stopped and the turntable 21 is stopped.

Step b. The pump continues to work as during normal drilling, i.e. a steady pumping cycle and the pump pressure is the same as during drilling. All other sources of interference such as those originating from generators, operation of cranes etc. are stopped. Wave motion and other noise sources are eliminated as far as possible in the case of offshore drilling.

Step c. As described above and shown in connection with fig. 5A, the coding down in the borehole is determined by a time control signal which is approximately synchronous with the time control signal on the earth's surface. Consequently, it is possible at the surface of the earth to detect when a single pulse is produced in the borehole and by knowing the transfer rate through the mud column, the exact time when the hydraulic pulse is received at the surface can be determined. It is thus possible to receive a single valve wave on the surface and know in advance the time when the valve wave will occur even if it is controlled by noise. In many cases, a single valve wave will stand out clearly from the noise so that visual observation on an oscilloscope is possible. The hydraulic transient produced by the valve 40 is received; thus off ■ the converter 51 at a known time.. Step d. The signal from the converter 51 passes the filter 150 for selective passage of frequencies in the range from 0.1 to 10 Hz. When the drilling itself is stopped, the ambient noise U(t) can be disregarded and consequently the signal appearing at the output of the filter 150 is of the form F(t) = B (t) + P (t).

Step e. The signal F(t) from the filter 150 passes a delay element 152, a subtractor 160 and an analog-to-digital converter 163 as described above. Usually during drilling, the signal from the converter 163 will have the form xfc = bfc + ut where ufc is noise due to drilling. However, since the drilling has stopped, the ambient noise ufc will also be imperceptible here. Under these conditions, x ~ bt- The signal xfc thus practically represents a noise-free signal which corresponds to the valve wave which represents the information-carrying signal.

Step f. The function xfc~ bfc is saved.

At the above steps, the output signal will

from the converter 163 be supplied to the element 17 3 for storage.

In terms of frequency, the memory function afc of the filter 170 can be expressed:

where f is the frequency and B* (f) is a Fourier transform of the signal bfc.

The elimination of ambient noise when it is white noise can in some cases be done by means of an autocorrelator instead of an adaptive filter such as the filter 170 of FIG. 9. In that case, fig. 9 is modified by eliminating the filter 174, the computer 172 and the element 173 and replacing it with an autocorrelator. The input terminals in the autocorrelator are then connected to the output wires 164 from the analog-to-digital converter 163. At the same time, the output wires from the autocorrelator are connected to the input wires 174 for the digital-to-analog converter 181. The output signal from the digital-to-analog converter can pass a delay element 190, an inverter 192, an AND - gate circuit 193 etc. as shown in fig. 9. In some cases, however, the output signal from the digital-to-analog converter can be fed directly to a recording device.

In fig. 9, a signal representing a function y(t) which is shown in fig. 10D and which has a shape similar to the double valve wave b(t), i.e.

y(t) ~ b(t) which represents a single bit in the digital signal that operates the valve 40. It is clear that such a function is not very applicable to represent a very short time interval corresponding to a single opening and closing of the valve 40. It is therefore necessary, as mentioned in step 3, to transform the double valve wave into a single short pulse that occurs simultaneously with the operation of the valve. This is achieved by means of a delay element 190 controlled by the timing generator 191 together with the inverter 192 and the AND gate circuit 193 (coincidence network) as shown in fig. 9. The delay element 190 receives on the conductor 192 the signal y(t) from the digital-to-analog converter 181. This delay element is controlled by the time control generator 191 so that a delay equal to the time interval T m is achieved on the output conductor 195. The delayed function b(t - Tm) is shown on fig. 10E and is supplied via the conductor 195 to the inverter 192 because on the conductor 197 from

element 192 and obtain an inverted delayed double valve wave expressed by -b(t - T m ) and which is shown in

fig. 10F.

The signal -b(t - T ) is supplied via conductor 197 to the AND gate circuit 193. At the same time, the signal b(t) is supplied

from the digital-to-analog converter 181 via conductors 182 and 200 to the AND gate circuit 193. Each signal b(t) and -b(t - T ) contains pulses having positive and negative polarity. By comparing the sigal in fig. 10D with the signal on

fig. 10F, it appears that there is only one pulse in fig. 10D which occurs simultaneously with the pulse in fig. 10F. This pulse occurs in the time interval from t^ to t^ in fig. 10D and from tg to t^g in fig. 10F. The times t^ and tg are coincident because t0 = t, + T and tQ = t, + T . On the same

3 1 m 3 9 1 m

way, the times t^ and t^^ are coincident because t^ = t-^ + T +T and t,. = t, + T +T. Accordingly, a is derived

n m 3 10 1 n m

single coincident pulse from the double valve wave b(t) as shown in fig. 10G. The AND gate circuit 193 which on the input conductors 200 and 197 receives signals representing the functions b(t) and -b(t - T ) consequently produces on the output conductor 210 a single pulse as shown in fig. 10G.

For the sake of simplicity, in this embodiment, only a single pulse is shown which is produced and which is mainly coincident with a single opening and closing of the valve. However, it should be noted that during actual drilling and simultaneous measurement, a coded series of single pulses representing a measurement performed by a selected sensor for a selected parameter will appear on the output conductor 210.

The coded sequence of single pulses on the output conductor 210 from the AND gate circuit 193 is supplied to a digital-to-analogue converter 211 which is controlled by a time pulse generator 212. On the output conductor 214 from the digital-to-analogue converter 211, a signal representing the measurement of the selected parameter is obtained in analogue form. This signal is recorded using a recording device 54.

Fig. 13 shows another embodiment of the invention where a cross-correlator can be used instead of the adaptation filter to eliminate noise. There is a close analogy between the treatment of the two functions given by equation (20a) and cross-correlation. Cross-correlating the two functions gives the same result as passing the first function through a filter whose memory function is a reversal of the second function. Fig. 13 shows how the same operations as with an adaptive filter can also be performed using a cross-correlator 200 which has two inputs 201 and 202 and an output 203. The signal x from the analog-to-digital converter 163 is fed to the input 201, while the signal bfc from the storage element 173 is applied to the input 202. A signal representing the cross-correlation of xfc and bfc then appears at the output 203. It is clear that the cross-correlated signal at the output

20 3 is identical to the signal y^ expressed by the equation (33)

and produced by the adaptation filter 170 in fig. 9. The cross-correlated signal is further processed as shown in fig. 13 in the same way as the signal from the adaptation filter 170 in fig. 9.

When the ambient noise is if noise, the auto-correlation q of the noise function is equal to zero for t ^ 0. Below we will treat the case that unwanted noise u has a known autocorrelation function q where the coefficient qfc is not necessarily zero for t f 0. This is the case for autocorrelated noise as opposed to pure white noise whose autocorrelation coefficient is qQ. A suitable form of an adaptive filter and associated components is shown in fig. 14. In this case it is necessary not only to store the information-carrying signal bfc as e.g. using the element 173, but also the noise signal u . On

fig. 14, two storage devices 17 3 and 224 are therefore arranged. The storage device 173 performs a function which is identical to the storage element in fig. 9 and 12. It serves to store the function bt. On the other hand is

the function of the storage element 224 to store the noise function ufc. Withdrawal of the functions bfc and ufc from the storage element 173 or 224 is supplied via channels 225 and 226 to a computer 228. The function of the computer is to transform the input signals from channels 225 and 226 into data to determine the memory function of the matching filter 220 and which is supplied via channel 230.

The process will now be the same as before, except that it must be taken into account that the noise ufc is no longer white noise. The adaptation filters that will be described here are independent of any amplification factor k which, for the sake of simplicity, is chosen equal to one.

The same definition of the signal-to-noise ratio u must be used here so that:

It is desirable to make u as large as possible by assuming that the noise signal ufc is an autocorrelated signal. It is appropriate to introduce matrix information at this point, namely:

which denotes the row vector (n + 1) which characterizes the memory in the adaptation filter 220. Furthermore, the row vector (n + 1) which defines the time reversal of the signal bfc denotes the autocorrelation matrix (n + 1) times (n + 1) of the noise signal. This can be expressed:

where (') denotes matrix transposition.

To make u as large as possible, the magnitude according to equation (46) must be differentiated against the filter vector a and the result is set equal to zero.

A relationship is then achieved:

which can be printed in the form:

This is the matrix form of a set of (n + 1) linear simultaneous equations with (n + 1) unknown filter coefficients (a^a^,... an). The solution of these gives the desired optimal matching filter in the case of autocorrelated noise. This method is very efficient, and it is therefore possible to calculate the matching filter of large length using the computer 228. The known quantities in the calculation are the noise auto-correlation matrix q and the time-reversed -t/single-wave signal b ^ , while the unknown quantities are the filter coefficients a^_. The filter coefficients represent the memory function of the adaptive filter 220.

The calculations necessary to determine the memory function of the matching filter 220 are performed by the computer 228. The computer receives from the storage elements 173 and 224 data relating to the functions b^ and u. Upon receipt of q^_, the noise autocorrelated matrix is calculated and upon receipt of bfc, the time-reversed of this signal is determined. The unknown filter components a are then calculated and transferred via the channel 23 0 to the adaptation filter 220.

The output signal from the matching filter 220 is supplied to a digital-to-analogue converter 181 and is further processed in the same way as the output signal from the matching filter 170 in fig. 9.

In the frequency range, the memory function of the adaptive filter 220 can be expressed as:

where B<*>(f) is the Fourier transform of the time-reversed signal b = (bQ, b-^, ...bn) and Q(f) which is the spectrum of the noise in the interval (f + df). The physical meaning of the equation (4 9) is simple. The larger the amplitude spectrum |B(f)| for the signal is and the smaller the density spectrum Q(f) for the noise is in the interval (f,f + df) the more the adaptation filter will let through frequencies in this interval. Thus, if the density spectrum Q(f)

for the noise is smaller in some intervals in the frequency band occupied by the signal, the less the adaptation filter will attenuate in this interval.

Storage of the signal b^_ in the element 173 takes place in the same way as in steps (a) to (f) in connection with fig. 12.

The storage of the noise signal ufc by means of the element 224 takes place in a different way. As previously mentioned in connection with fig. 12, it is possible to receive and store a noise-free signal. As a result of synchronism between the time control on the surface and down in the borehole, it is likewise possible to receive and store signalless noise, i.e. signal received by the converter 51 during normal drilling and which contains all the noise due to the drilling operation, but nothing carrying information noise. Also in this case, fig. 12 is used to illustrate the method. The steps to achieve a registration of the function u(t) can be as follows: Step ( a \. The weight of the drill bit is applied and normal drilling operation is carried out.

Step ($) . A period of time is selected when no information-carrying signal appears, e.g. a pause between binary words. Step (y). A signal is obtained that represents the pressure variation in the drilling mud in the converter 51. This signal passes the filter 150. Since the signal b(t) in the time period selected in step ((3)) does not occur, the signal from the filter 150 has the form F (t) = P(t) + U(t).

Step (6). The pump noise signal P(t) is eliminated. This takes place with the help of the delay element 152 and the subtractor 160. The resulting signal is supplied to the analogue-to-digital converter 163. As no information-carrying signal appears,

bfc = 0, the signal from the output of the analog-to-digital converter 163 will have the form xfc = ufc.

Step (e). A registration of the function x^_ = ufc is achieved by using the storage element 224 in the output of the analog-to-digital converter 163 as shown in fig. 12.

From the above it appears that if the noise is white, the adaptation filter 17 0 and the associated components as shown in fig. 9 optimal value of the signal-to-noise ratio y. If the noise is not white, but has a known autocorrelation function, the adaptive filter 220 and the associated components in fig. 14 ensure optimal value of p.

Fig. 15 shows part of the equipment on the ground surface, comprising a filter which works differently from the matching filter of fig. 9 or 14. The adaptation filter of fig. 9 or 14 gives optimal signal-noise protection because it is a linear filter. The filter 240 on

fig. 15, on the other hand, is a pulse-shaping filter or a Wiener filter that provides an optimal noise ratio in that it is a linear filter that makes the square of the difference between the desired output signal and the actual output signal as small as possible.

The pulse shaping filter 240 in FIG. 15 receives via the input channel data relating to the function xfc = b^_ + ufc derived from the analog-to-digital converter 163. The pulse-shaping filter with a length (m + 1) has a memory:

which transforms at least the input signal x^_ = (Xq,x^,... x n) with a length (n + 1) into an output signal (m+n+1) with a length z = (zQ,z^, ...zm+n)• A function diagram for such a filter is shown in fig. 16. Three signals appear here, namely the input signal xfc, the relevant output signal zfc and the desired output signal bfc. The signal b is a double valve wave as shown in fig. 10C.

The output signal z is an envelope of the filter memory function ffc with the input function xfc, i.e.:

The problem is to determine the memory function

ffc so that the relevant output signal zfc comes as close as possible to the desired output signal b^. For the selection of the memory function, the following quantities must be made as small as possible: I = the sum of the square of the error between the desired output signal and the filtered valve wave signal + the value v

of the filtered noise,

where v is a predetermined weighing parameter.

The calculations necessary to reduce I are carried out in the computer 245 which has three input channels 246, 247 and 248. The storage element 173 delivers to the computer 245 via the channel 246 data relating to the function bfc. The storage element 224 transfers in the same way to the computer 245 via the channel 247 data relating to the function u.

Channel 248 is used for transmission to computer 245

data relating to the function x which is also supplied to the input conductor 241 in the pulse-shaped filter 240.

When receiving the input signals bfc,out and x via channels 246,247 resp. 28, the computer 245 will perform certain calculations to be described in more detail below and via the channel 251 deliver the desired data relating to the memory function of the filter 240 to the pulse shaping filter 240. The relevant filter output signal zfc is at least in terms of error as close as possible to desired output signal b^, in other words:

as shown in fig. 10D.

The operations carried out by the computer 245 will be considered in detail below. With symbols, the quantity I can be expressed:

where E indicates the mean value and where: represents the filtered noise. Simplifying the expression for I yields:

where t is a dummy time index and where q is the autocorrelation

of the received noise. Differentiating the expression for I with respect to each of the filter coefficients and setting the derivatives to zero yields a set of simultaneous equations which are

for t = l,2,...m. In the equations above, the quantities rt, -s ,qt, -s and g^ t are known, while the quantities of f s are unknown,

Calculations performed by the computer 245 serve to determine the parameters r- t- s, c* t- s and gt ^ra inn9an9~s~ the functions supplied on the channels 248, 247 and 246, and then solve the equations (56) for the unknown quantities f.g. The parameters rt_s»»v» °9 9 4- are ^e^ined as follows: The parameter rfc_s is the autocorrelation of the input signal x^_ supplied to the computer 245 via the channel 248. The parameter q is the autocorrelation of the noise signal u^ supplied to the computer 245 via the channel 247. The parameters g are defined as the cross product coefficients between the desired output signal bfc and the input signal xfc, so that:

for t=0,l,2,...m. In the expression for g, the desired output signal bg is supplied to the computer 245 via channel 246 and the input signal xfc is supplied via channel 248. The parameter v is a weighing parameter which is given a suitable value as will be described in more detail below.

The parameters r+-_s'c34-_s °9 g. 13631:611111163 thus

by the computer 245 and then this solves the equations and delivers on the output channel 251 the quantities f which are supplied to the memory in the pulse-forming filter 240. The output signal

z from the filter 240 is, at least in terms of energy, as close as possible to the desired output signal bfc.

Then the rna trick of the equation, namely [rt_s + V(3-(-_^

has the form of an autocorrelation matrix, the equations can be solved efficiently by the iteration method.

It should be noted that the processing time required to solve the simultaneous equations given above

2

in a filter with m coefficients is proportional to m for the iteration method compared to m 3 for the usual method of solving simultaneous equations. Another advantage of using the iteration method is that it only requires computer storage space proportional to m instead

2

for m as is the case with normal methods.

In the construction of a pulse-shaping filter, two conditions must be taken into account, namely (a) as closely as possible to shape the function z^ to the desired function bfc. (b) to produce as little output energy as possible when unwanted stationary noise is the only input signal.

In many practical cases, a filter is required to meet both of the above-mentioned conditions at the same time, and one is therefore faced with the problem of finding a suitable compromise between the two conditions. Therefore, a suitable value of the parameter v is chosen as a mutual weighing between the two conditions.

There are situations where the parameter v is zero. In this case, equation (53) takes a simpler form,

in fact:

and the computer 245 does not need data representing out- In this case, the storage element 224 of FIG. 15 is looped and the computer 245 only needs to have two input channels, namely 246 and 248.

It appears here that the operation of a pulse-forming filter and an adaptation filter are not exactly the same, i.e.

for a given input signal, the output signal from these filters are not identical. The expression yfc ~ bfc which can be used for a matching filter has been used above to indicate that the signal expressed by y representing the output signal from a matching filter has a close approximation to the double valve wave bfc. Consequently, it was pointed out that the same curve in fig. 10D represents the functions y(t) as well as the function b(t). It should be noted that the expression zfc ~ bfc applicable to a pulse-shaping filter is above used to indicate that the signal expressed at 7. representing the output of a pulse-shaping filter closely approximates the double valve wave bfc. Consequently, the same curve in fig. 10D represent the function z(t) as well as the function b(t). Strictly speaking, the same curve cannot be used to represent the functions b(t), y(t) and z(t). However, since both y(t) and z(t) are close approximations to b(t), it can be assumed to be sufficient for the sake of explanation to use the same curve for explaining the function of an adaptation filter as a pulse shaping filter.

In fig. 17, the output conductor 162 from the subtractor 160 has both the information-carrying signal b(t) and the noise signal u(t). The signal b(t) is the double valve wave shown in fig. 10C. The mixture of the signal b(t) and u(t) is supplied to the analog-to-digital converter 163 which emits the converted digital signals bfc and v on the output conductor 164,

which corresponds to the analogue signals b(t) and u(t). The mixture of these two signals is supplied to the input conductor 300 in a peak-forming filter 351. This filter in the present case has the form of a Wiener filter where the desired shape is simply a peak.

As mentioned above, a double valve wave b(t) as shown in fig. 10C of two half valve waves, i.e.

the half valve wave "A" and the half valve wave "B". The half-wave "B" follows the half-wave "A" after a time T .

The operation of the peak forming filter used in fig. 17 is to transform the two half-waves into respective clear peaks. A double valve wave bfc is therefore transformed by means of the peak forming filter 351 into a pair of peaks.

fig. 18A to 18F show six possible positions of a pair of peaks M^ and N, relative to the double valve wave applied to the input 300 of the filter 351.

In all figures 18A to 18F, T k is the time interval between the peaks M and N^. is the point where the peak M^ intersects the abscissa expressed in milliseconds. The distance OH^ represents the time position of the peaks in relation

to the double valve wave. In fig. 18A is thus the time delay OH^ = 0, i.e. that the first peak M^ is located at the beginning of the double valve wave. The five curves in fig. 18B to 18F show an increasing value of the time delay OH^. One of these curves represents the maximum value of time delay for which the resolving power of the filter is greatest. For this time delay, the output signal from the filter is significantly sharper than for any other time delay. The way to obtain the optimal value of the time delay, the optimal length of the memory in the filter and the optimal value of the time interval T will be described in more detail below.

The two peaks at the output from the filter 351 represent a single bit in the digital signal that operates the valve 40. As pointed out in connection with fig. 9 desirable to transform the two peaks into a single peak or pulse. A system similar to that in fig. can be used here. 9. A delay element 303 controlled by a time pulse generator 304 is therefore used in combination with an inverter 306 and an AND gate circuit 307 which serves as a coincidence circuit. These components are arranged in the same way as in fig. 9. However, the degree of delay is in fig. 17 differs from fig. 9. On fig. 17

the delay element 303 must produce an output signal which is delayed in relation to the input signal by the time T, , while in fig. 9 is the delay in the delay element

193 equal to T .

m

A coded series of single pulses appears at the output of the AND gate circuit 307 and these are supplied to a digital-to-analogue converter 308 which is controlled by a timing control generator 309. At the output of the digital-to-analogue converter 308 a signal appears in analogue form that represents the value of the selected parameter in the borehole. This signal is recorded on a recording device 54.

In some cases depending on the electronic circuits selected in the various components of fig. 17, the inverter 306 is not required because its function can be performed by suitable design of the AND gate circuit.

Fig. 19 shows an alternative device for noise elimination by means of a peak-forming filter. In fig. 17, special aids were used to eliminate pump noise, namely the delay element 152 in combination with the subtractor 160. In fig. 19, the noise elimination is simplified in that the delay element 152 and the subtractor 160 are looped. In fig. 19, the signal F(t) on the output 151 of the filter 150 is digitized by means of an analog-to-digital converter 350 and then supplied to the peak-forming filter 351a, which is designed differently from the peak-forming filter 351. In fig. 17, this filter is intended to convert a double valve wave such as e.g. shown on

fig. 18A to 18F to a pair of peaks separated by an interval Tfc. The filter 351a in fig. 19, on the other hand, is intended to convert a single valve wave into a single peak. Different positions of this peak in relation to a single valve wave are shown in fig. 20A to 20F.

It should be remembered that each individual valve wave

which is supplied to the input of the filter 351a and each individual peak delivered from the filter represents a single bit in the digital signals operating the valve 40. The coded sequence of peaks in the digital signal from the filter 351a is supplied to a digital-to-analog converter 352 where they are converted into a coded sequence of peaks where each peak represents a single bit of the information encoded by

the measuring equipment down in the borehole. The series of these bits represents in digital form the measurement of the selected parameter. However, it is necessary for recording and/or indication to reproduce the measurements in analogue form. Accordingly, the signal from the digital-to-analog converter 352 is supplied to a digital-to-analog converter 36 2 which supplies a signal with a value representing the measurement of the said parameter. This signal is recorded using a recording device 54.

It should be noted that the transformation of the double valve wave into two peaks by means of the filter 351 on

fig. 17 or the conversion of a single valve wave into a single peak by means of the filter 351a of fig. 19 can only be an approximation. A pure peak, i.e. a delta function cannot be achieved. However, the purpose of the invention is to increase resolution, i.e. to achieve an output signal that is significantly sharper than the input signal.

Theoretically, a peak-forming filter could be calculated exactly if the filter's memory function were infinitely long. For an accurate filter it is also generally necessary to delay the desired peaks infinitely relative to the input signal. In practice, however, it is necessary to have a filter with a finite memory function so that at best only an approximation can be achieved.

In practice, the filter's memory function can

be of the order of magnitude of the duration of the input signal. It must be assumed that it is possible to place the desired one

top in any selected position, e.g. as shown in the six positions in fig. 18A to 18F for the filter 301 of FIG. 17 and similarly to fig. 20A to 20F for the filter 351 of FIG. 19. The optimal position of the peaks is determined for each of these cases. It should be noted that the position of the peak is an important factor in obtaining the best approximation of the output signal to the desired peak.

In the present case, the peak-forming filter is also a pulse-forming filter of the Wiener type and the conditions for its construction are analogous to the

as described above.

To determine the optimal value of the time delay and the optimal length of the memory function for the filter 301 of FIG. 17, it is necessary to plot a double valve wave bfc which is the digital version of b(t). The necessary steps for obtaining such a drawing are indicated by (a) to (f) above with reference to fig. 12. The recorded signal b^ is stored in the storage device 173 in fig. 12. In order to determine the optimum value of the time delay and the optimum length of the memory function for the filter 351a, it is necessary to make a plot of a single valve wave B^

which is the digital version of B(t).

The different positions of the top from the filter 3 51a in fig. 19 corresponding to the different delays as shown in fig. 20A to 20F, can be expressed:

(1,0,0, 0,0): Delay = 0

(0,1,0, 1,0): Delay = m + n - 1.

(0.0 0.1): Delay = m + n.

The operation of a peak forming filter corresponds to the various delays shown in the functional diagrams 21A,

21B and 21C. In all these cases the input valve waveform is the same Bfc recorded and stored as explained above. The desired output signal in fig. 21A is a peak (1,0,0) i.e. no delay. The corresponding memory function for zero delay is F° = (F^F^jF^, F°) and the real output signal is W° = (W°,W°, W°). The same applies to fig. 21B and 21C. Each position of the peak corresponds to an energy error E. The normalized minimum of energy error E represents a very appropriate way of measuring the operation of a pulse-forming filter of the Wiener type and in particular a peak-forming filter. When the filter works perfectly, E=0, which means that the desired signal and the actual output signal agree for all values of the time. On the other hand, the case E = 1 corresponds to the worst result, i.e. that there is none

agreement between the desired signal and the actual signal. Instead of the size E, it is desirable to use the complement of E, which is to be designated as the filter's operating parameter P, so that:

A perfect filter effect will therefore be achieved when P = 1, and the worst effect when P = 0.

Fig. 22 shows the procedure for measuring the parameter P. A computer 400 has three input signals 401, 402 and 404. The channel 401 receives from the storage element 403 data representing a half valve wave B^, the channel 402 receives from the time delay element 405 data concerning peaks with different time delays, and the channel 404 receives data from the memory duration controller 406 concerning peaks with different memory durations. The output 410 from the computer 400 is connected to a measuring instrument 411 for measuring the parameter

P.

For constant filter duration, it must be assumed that there is at least one value for the time delay where P is the largest possible. Fig. 23 shows a curve for P as a function of the time delay for several filters with a fixed duration. The highest point on the curve M.^ corresponds to a time delay ON-^ and the selection of this point in time leads to an optimal time delay for the filter. It should be noted that the curve in fig. 23 applies to filters with a fixed duration.

It also shows what happens if the filter's memory duration is increased with a constant time delay.

Fig. 2 4 shows a curve for the parameter P as a function of the length of the filter for a desired and a fixed time delay. It is clear from this curve that it approaches asymptotically

to the largest value of P the longer the filter is. The curves in fig. 23 and 24 are obtained by means of the device on

fig. 22.

The two important criteria are the filter's delay time and the memory duration. One can always improve the filter by increasing the duration of the memory function, but in practice this cannot be made infinitely long. On the other hand, one can strive for the desired output signal to have a time delay that leads to the largest value of the parameter P for a selected filter duration. This time delay in the filter is not in any way a disadvantage and can significantly improve the filter.

The parameter P as a function of the time delay and a constant duration as shown in fig..23 or the parameter P as a function of the filter's memory duration for a constant time delay as shown in fig. 24, can be of great help, but does not tell the whole story. The ideal would be to investigate the dependence of P on the time delay and memory duration for all practical values of these variables. One way to achieve this is to plot P by taking the time delay as the ordinate and the memory duration as the abscissa. A group of values P can then be drawn up and from these it can be seen which combination of time delay and memory duration gives the optimal filter effect. This is shown in fig. 25 for three parameters P]_'P2 Q<3 P3 • What we are most interested in is the largest value of P for the best possible filter effect. Fig. 25 makes it possible to choose the best combination of the filter's time delay and the memory duration.

Even if the measuring equipment is equipped for binary coding, other coding methods can sometimes also be suitable.

For example, for a gamma ray sensor or an electronic compass inclinometer, pulse time code may be preferable.

In some cases, particularly where several numbers are to be sent in turn, pulse time coding has advantages. In the case of an electronic compass, it is necessary to send five numbers in order to measure a magnetic course. By using pulse time coding, significant savings can be made both with regard to the energy required for the battery and the time it takes to transfer data.

Conventional pulse timing coding is shown in fig. 26A e.g. for passing the value of three parameters. A number of voltage pulses are transmitted and the duration t^,t2 and t^ for the individual pulses a,b and c represent the size of the parameter to be transmitted. It should be noted that between each pulse a pause is necessary to

separate the pulses from each other. In fig. 26A, the pulses a,b and c are analog with three binary words each separated from each other by a time interval T w. These pauses naturally seem unfavorable for fast data transmission because the pauses themselves do not contain any information. Long duration of the pulses is also unfortunate for the remote measurement system.

In the present case, pulse time coding is used as shown in fig. 2 6B where it is not the duration of the pulse that is a measure of the parameter, but the time between two consecutive very short pulses.

Instead of transmitting long pulses of variable duration, short pulses of constant duration are used, e.g. a few milliseconds and the time between the pulses is a measure of the size of the parameter. In this way, no extra time is needed to separate the pulses from each other. In fig. 26B, parameter No. 1 is represented at time t^ between pulse Pq and P^. The parameter no.2

is represented at time t2 between pulse P^ and P2, and parameter No. 3 is represented at time t^ between pulse P2 and P^. The pulse P-^ represents the end of the time interval t^ and also the beginning of the time interval t2, the pulse P2 represents the end of the time interval t2

and also the beginning of the time interval t^ etc. There is therefore no time loss in this case between each important time interval, i.e. that T w in fig. 26A is zero.

By using the pulses P^, P2 and P^ both to indicate the end of a time interval and the beginning of the next time interval, the time loss becomes zero and all time is used for the transmission of data. In binary coding, each word is immediately followed by the next word. Only at the end of a series of words is there a pause T. In the next series, the time intervals pg'Pl'P2'P3 will usually be somewhat different for the data represented by the times tl't2't3 vanli9v:i-3 vary and each series of data represents e.g. a new condition in the borehole.

Fig. 30 shows a device for pulse time coding according to the invention. Practical measuring equipment for boreholes naturally uses modern electronic integrated circuits, but for the sake of simplicity, fig. 30 a mechanical step switch and a simple mechanical relay. The sensors 1,2 and 3 are connected to corresponding contacts in the selector 285 which is operated by an electromagnetic device with a winding 286 from the step selector in the "O" position. The battery 288 will then charge a capacitor 290 via a resistor 289

with a current determined by the resistor 289, the capacity of the capacitor 290 and the voltage of the battery 288. A trigger 291 produces a single electrical pulse when the voltage supplying it exceeds a predetermined voltage. The output signal from the trigger 291 energizes the winding 286 and the movable contact 287 in the tap-changer 285 moves to the next contact. At the same time that the trigger 291 energizes the relay 292, the capacitor 290 discharges to ground.

When the movable contact 287 is moved to the next position, the process is repeated with the exception that instead of the voltage from the battery 288, the output voltage from sensor No. 1 is connected to the circuit and the pulse P^ is produced at the time when the capacitor is again charged to the trigger voltage for the trigger 291. This time is

RC

proportional to the value ( —^ ) where R is the value of the resistor 289, C is the capacity of the capacitor 290 and Vg is the output voltage from the sensor. The time t^ is inversely proportional to the output voltage from the sensor.

After activation of the trigger 291 by the voltage from sensor No. 1, the process is repeated and when the voltage on the capacitor 290 reaches the trigger voltage, the trigger 291 produces a sharp pulse which energizes the relay 292, discharges the capacitor 290 and energizes the tap-changer 285 so that the movable contact 287 moves to next con-

tact.

The step coupler 285 is fed forward in turn and connects the sensors 1,2 and 3 with the resistor 289. Pulses produced by the trigger 291 when the movable contact 287 is in the "0" position corresponds to the pulse PQ in fig. 26B and the pulse produced by the trigger when the movable contact 287 is in positions which connect the sensors 1,2 resp. 3, pulses ,and P^ are produced. The associated time intervals t^,t2,t3 represent inversely proportionally the voltage from the individual sensors.

Curves 26A and 26B show the principle of pulse time coding which can be used down the borehole instead of the analog-to-digital converter 102 in fig. 4A. The decoding on the surface is carried out by ordinary pulse time coding and shall not be described in more detail here.

In fig. 26C represents TPq,TP-^,TP2,TP^ etc.

a series of pulses received at the detection site on the Earth's surface. These pulses occur at times TqjT-^,

T2, T3 etc. In the case of the pulse time coding described in connection with fig. 26B, the time between successive pulses is used to indicate the magnitude of a parameter. If 3 parameters are to be measured remotely, the code can be as shown in fig. 26C where: ^ - Tg is a time interval representing

the No. 1 parameter.

T2~ Tl is a time interval representing

the #2 parameter.

T3~ T2 is a time interval that represents

The No. 3 parameter.

When measuring mud pulses during drilling, it is sometimes necessary to measure with great precision. As the speed of sound in the mud column is not always constant and noise and attenuation vary, the time intervals between pulses received on the earth's surface cannot exactly correspond to the time interval between the pulses produced down in the borehole. In other words, there is often an uncertainty on the Earth's surface when it comes to the exact time

for the reception of a specific pulse.

It shall be assumed that the absolute uncertainty for the time at which each pulse is detected is plus or minus 0.2 seconds or a total of 0.4 seconds. To achieve an accuracy of + 1% for T, - TQ with a total absolute error of 0.4 seconds, the time between pulses must be at least 0.4 x 100 or 40 seconds. Furthermore, as the apparatus does not always produce a clear, sharp pulse, at least two detections are necessary. If both detections give the same result, the data is confirmed. Consequently, to achieve the desired accuracy of + 1%, 80-120 seconds will be required per measured parameter, i.e. approx. 2 minutes per parameter. ,

With the improved pulse time coding according to the invention, much greater accuracy is also achieved. For each pulse <p>ø'Pl'P2'P3 not only a single mud pressure pulse is sent, but a group of at least three mud pressure pulses located at different distances as shown in fig. 26D which will be designated below as a triple group. The time interval in each triple group can be t^ which is the time from the first pulse to the second pulse, t2 which is the time from the second pulse to the third pulse and t^ which is the time from the first pulse to the third pulse. In this case, Tq also represents the time of arrival of the triple group TPq, the arrival time of the triple group TP^ T^ the arrival time of the triple group TP2 and T, the arrival time of the triple group TP^, and

- Tq is the time interval that represents the parameter

No. 1.

T2 - is the time interval representing the parameter.

No. 2.

T3~<T>2 is the time interval representing the parameter.

No. 3.

The advantage of this system is that if a momentary error occurs which causes a pulse in the group not to be received, the error can be instantly detected because a triple group contains two pulses instead of three. Furthermore, since the time intervals t^,t2,t^ are different and known, it can be determined which pulse in the group is missing and finally, because t^,t2ft^ are known, the correct correction can be provided to determine the time intervals - Tq,

- T2, T2 - with the same accuracy as if all the pulses had been present in the triple group. The triple group also has a further advantage, namely because it is difficult to determine the exact arrival time of a particular pulse, the triple group enables a significantly better determination of the arrival time. It can e.g. take the arithmetic mean value of the arrival times for each pulse in the triple group or by using modern computer technology and record even better accuracy for the arrival time. A

Fig. 29 shows the electronic logic system which produces the triple group pulses shown in Fig. 26D. A sensor 101 produces an electrical voltage that indicates the magnitude of the parameter down in the borehole. This voltage is supplied to a voltage-controlled oscillator 601, which is connected to a scaler 602 which in turn is connected to a trigger 603 which produces a series of electrical pulses which are separated by time intervals representing the magnitude of the voltage from the sensor 101. The time interval between the pulses Pq and P^ on fig. 26B is therefore a measure of a parameter measured by one of the sensors 101 in FIG. 4A.

The part of fig. 29 which is framed with

dashed lines show the circuits 607, 608 and 609 which generate the triple group pulses. These circuits are electronic monostable multivibrators which, in accordance with the pulse from the trigger 603, produce a single pulse of permanent

called D3_#D2 or D3 as shown in fig. 29. The circuit; 610 outputs a signal that is proportional to the first time derivative of the input signal. Output signal

G,H and I which are pulses of opposite polarity are separated by the time intervals Dj_rD2 0<? D3 • Rectifiers 611 only let through positive pulses which are supplied in parallel to the conductor 612 which results in a signal J which is the desired signal as shown in fig. 26D. Each single pulse produced by the trigger 603 will therefore produce three pulses and different time intervals in the form of a triple group J. In practice, the interval D is made very short compared to D2 and D3 and should only be a few microseconds, while D2 and D-, are intervals from a few milliseconds to several hundred milliseconds. When analyzing the course, one can therefore assume = 0.

In the output signal J in fig. 29, the pulse P-^ therefore indicates the end of the output pulse from 6 07 which in practice also means the beginning of the output pulse because the pulse length is assumed to be zero, the pulse P2 denotes the end of the output pulse from 608, and the pulse P3 denotes the end of the output pulse from 609. I and since Dj_ can be considered to be zero, the time interval t-^ = D2, the time interval t^ = and the time interval t2 = D-^-D., will therefore. The part of fig. 29 which is framed by dashed lines emits triple groups J at the output 612 in response to a single pulse applied to the input.

The device in fig. 29 can on fig. 4A is added

in between a selected sensor 101 and the drive source 104. In other words, when pulse time coding according to fig. 29 is used, the analog-to-digital converter 102 and the processor 103 are looped because they are designed for binary coding, and the energy source 104 is driven directly by the output of the amplifier 613 on

fig. 29.

When pulse time coding with triple groups is used

instead of binary coding, it is necessary to decode the triple group signals on the earth's surface. In fig. 9,12, 13,17 and 19, the signals sent down from the borehole are assumed to be binary signals. To change the system for receiving signals which are pulse time coded with triple groups as shown in fig. 29 and 26D, it is necessary to arrange one

specific decoder such as shown in fig. 27 between the filter 150 and the subsequent apparatus. For this purpose, the conductor 151 in fig. 9,12,13,17 and 19 broken and a suitable decoder introduced. In some cases, it is more desirable to place the decoder between the subtractor 160 and the analog-to-digital converter 163 in the conductor 162.

In fig. 27, a selection circuit 316 is arranged which produces a single output pulse in response to a triple group. This pulse is supplied to a time-amplitude converter 317, i.e. an electronic circuit which produces a direct voltage on the conductor 319 which is a predetermined function of the time between two input pulses applied to the input from the conductor 318. The output signal from the converter is supplied to an analog-to-digital converter 320.

Fig. 28A shows the selection circuit 316 in detail. The input signal is applied in parallel to monostable multivibrators 321, 322 and 323 which produce a single output pulse of predetermined duration in response to the input pulse. The multivibrator 3 21 produces a long pulse of duration 13, the multivibrator 322 produces a pulse of shorter duration 12 and the multivibrator 323 delivers a pulse of even shorter duration 1^ as shown in fig. 28A. The signal from the multivibrators is fed to circuits that produce an output signal that is proportional to the first time derivative of the input signal. The output signal from these circuits is supplied to inverters 325 which produce an output signal which is a reproduction of the input signal but with the opposite sign as shown in fig. 28A. The output signals from the inverters are supplied to rectifiers 326 which only let through the positive pulse 326a, 326b or 326c. These pulses are fed to coin side circuits 327 or AND gate circuits. Each coin side circuit 327 delivers an output pulse on output c only when a pulse on input a and a pulse on input b occur simultaneously. The output signals from all three coincidence circuits 327 are supplied via the conductor 329 to a time-amplitude converter 317. The converter 317 supplies a DC voltage which is a predetermined function of the time between two successive input pulses. The output signal from the converter 317 is supplied to an analogue-to-digital converter 320 which converts the direct voltage into binary coded pulses.

The selection circuit, however, requires a more detailed description. The pulses P^, P2, P3 which are delivered by the amplifier 613 in fig. 29 is supplied to the drive circuit 104 in fig. 4A and is transmitted to the soil surface as mud pressure pulses by means of the valve 40. On the soil surface, these mud pressure pulses are detected, e.g. of elements on fig. 9 which comprises the converter 51, the filter 151, the delay element 152 and the subtractor 160. The pulses appearing on the conductor 162 in fig. 9,12,13 or 17 shall be denoted TP^ TP2 and TP-j corresponding to the pulses P1'P2 and P3 which are produced down the borehole.

Fig. 28B, 28C, 28D, 28E show the reaction on fig. 28 of the pulses TP17TP2 and TP^. When the pulse TP-^ occurs on conductor 151 or 162 in fig. 28A, all three multivibrators 321, 322 and 323 will be triggered and each of them will produce output pulses with their own characteristic and determined and fixed length 1^, 1^ and 1^. When the pulse TP^ triggers the mutlivibrators, output voltages A^,B^,C^ are produced as shown in fig. 28B.

When the pulse TP2 is supplied to the multivibrator 321 it is not triggered because it is already in the triggered state. However, the pulse TP2 triggers the multivibrators 322 and 323 because they are not yet triggered,

and these produce the pulses B2 and C2 as shown in fig. 28B. When the pulse P^ occurs, the multivibrators 321 and 322 are not triggered because they are already triggered. However, the pulse TP^ triggers the multivibrator 3 23 because it has already returned to the initial state and it produces the output pulse C, as shown in Fig. 28B.

The time intervals 13, 12 and 1^ from the multivibrators 321, 322 and 323 are proportioned so that they correspond to the time delays resulting from the multivibrators 609, 608 and 607 in fig. 29 and consequently the end of the group iv aisse puiser will appear simultaneously and aen or open the uo-pori;-crets in fig. 28A. Fig. 28B shows the conditions when all the pulses TP.^, rP.sub.2 and TP.sub.3 occur.

Fig. 28C shows the same conditions as fig. 28B,

nen one of the pulses is missing, namely TP^.

Fig. 28D shows the same conditions, but here the pulse TP2 is missing and in fig. 28E lacks the pulse TP3. It should be noted that regardless of which pulse is missing, the output pulses from two of the multivibrators will always be terminated at time T. This is used to always produce at least two pulses simultaneously at time T regardless of which pulse is missing. As long as at least two pulses in the group are detected, the time of appearance of the output pulse of 329 in fig. 28A be the same. The pulse 328a in fig. 2 8A is produced when a group of pulses is received at the earth's surface Dg pulse 328a will occur when any two pulses in the group are detected at the surface.

The time-amplitude converter 317 in fig. 28A produces a DC voltage which has a predetermined functional relationship to the time between consecutive pulses 3 28a. An analog-to-digital converter 3 20 converts the amplitude of the DC voltage into a binary word. The binary Drds follow one another in rapid succession determined by the characteristics of the converter 3 20 and the associated timing generator.

It thus appears that the device in fig. 28A converts pulse timing coding with triple groups to binary code and the circuits following conductor 151 or L62 of FIG. 9,12,13 and 17 will function in exactly the same way as if the data was originally transmitted in binary Geodet form from the borehole.

In order to achieve shock waves as described above, there are certain limits for K^ r ie the degree of change of the opening of the valve, and ^ which is the time for open flow2. Experiments have shown that K z9 must be at least 5 cm/sec. and preferably lie within the range 20-150 cm<2>/sec. Tfa^ must be no more than 500 milliseconds and

preferably in the range of 50-150 milliseconds.

Furthermore, the synchronization pulses from the time control generator 155 must either be connected to the pump shaft or with a phase-locked loop. Other aids to achieve a control pulse frequency synchronous with the pump can of course be arranged. E.g. a known pump stroke counter can be used which is usually used in connection with the pump rod and which produces an electrical pulse per pump strokes. The time between such consecutive pulses can be divided into a sufficient e.g. 512 or 1024 at equal time intervals using a microprocessor or a phase-locked loop or some other means. In those cases where the pump rod is not available, a timing control frequency similar to that from the generator 155 can be produced by a microprocessor.

At shallow depths, it may happen that the valve waves are not formed correctly. In such cases, it is necessary to have a sufficient volume of mud flowing in the drill stem and sufficient hydrostatic pressure near the valve. It is clear that the invention is not only limited to special valve waves as shown, but can be used for other forms of pressure pulses that can be detected on the earth's surface as a result of the operation of the valve 40.

Various digital filters such as adaptive filters, pulse forming filters and peak forming filters are described above in detail. In particular, the mode of operation is that each digital filter is described clearly by a detailed description of the mode of operation. It is clear that by using modern computer technology, necessary programs can be provided on the basis of the description in connection with fig. 9,12,13,14, 16,17,19,20.

Different digital filters that have been described can also have other forms of processing measurements in the case of sludge pulses using valves of a different type than the valve described above. The other forms may include controlled throttling of the sludge flow in the main sludge flow. The digital filters described above can be used for any type of remote measurement system during drilling and other forms of monitoring where the drilling equipment is removed to lower the measuring equipment into the borehole. It can be used for the remote measurement system that uses pulses that represent any form of energy, such as e.g. electromagnetic acoustic or other pulse energy.

The pulse time encoding with triple pulse group as described above can also be used for acoustic monitoring in boreholes to achieve a very effective method for automatic correction of errors due to the absence of a pulse in the measurement during the transmission of the acoustic waves.

Pulse time coding with triple pulse groups has a very wide application area outside borehole monitoring. It can be used in any communication system for the transmission of messages from a sending station to a receiving station as well as different types of acoustic monitoring other than monitoring during drilling.

It is clear that in order to record and store a valve wave for later use in digital filters as described above, certain precautions must be taken as described above. It is sometimes desirable to derive a single valve wave rather than a double valve wave as in fig. 19 using a peak forming filter 351A. For derivation of a single valve wave, it is appropriate to synchronize the generation of the signal that is generated down in the borehole with the detection on the earth's surface. This can be done by replacing one of the sensors 1,2,3 and 4 down in the drill hole in fig. 4A with a time signal generator which will cause an even time interval between the operations of the pump 40 in fig. 4A. The way it works is then as follows: a) by stopping and starting the mud pumps on the ground surface in the correct order, as the switch 91 in fig. 4A is connected to the modified sensor, i.e. the generator receives evenly distributed pulses. Then a series of pulses will be generated by the valve at known times. Naturally, there must be a correction for the propagation time of the pulse from the borehole to the earth's surface. b) The equipment on the ground surface is controlled by a separate time control generator which is synchronous with the time and phase of the signal transmitter down in the borehole. c) With a suitable connection on the earth's surface, the discharge and storage of the double valve wave can be interrupted so that

the storage device only connects at the time of one valve wave, and is automatically disconnected during the subsequent second valve wave. Naturally, the same operation can be performed manually. This is easy when the valve wave is clear and clearly drowns out the noise. When the valve wave is drowned in noise, the automatic system described above can be used.

There are two interfering noise signals that obscure the reception of a usable signal B(t) (see equation 22). One is represented by the pump noise P(t) and the other represented by noise U(t) which is due to various drilling operations other than the pump noise. To eliminate these interfering signals, three filter systems have been proposed.

The first of these is the analog filter 150 to suppress the continuous component from the transducer representing the pressure produced by the pump 27 and other frequencies outside the range of interest.

The second comprises a delay element 152 and a subtractor 160 to suppress or eliminate the pump noise P(t).

The third comprises a correlator or a digital filter which can be an adaptive filter, a pulse shaping filter or a peak forming filter and also contain various elements such as storage elements and computers to determine the optimal value of the memory elements for the corresponding digital filters (see Fig. 9, 12,13,14 and 15). The purpose of this third system is to eliminate or suppress noise U(t).

These three filtration systems are connected

in cascade. In the embodiments described above, the first filter system is connected to the pressure converter 51,

the second is connected to the output conductor 151 and the third is connected to the output conductor 164 of the second system.

Each of the above mentioned filter systems are linear systems. The operation of these systems can therefore be changed or reversed. In the first row, the first filter can come first, and the second and third filter systems can be changed. In some cases, it may be necessary to use all three filter systems. In other cases, two may be sufficient, and in some cases only one of the filter systems. The system between conductor 182 and 210 can sometimes be eliminated, and a digital-to-analog converter 211 can be arranged to receive the double valve wave.

When the signal is produced as in step (f) above and stored, it can be cross-correlated with the signal produced by the converter 51 and the pre-processed signal on the conductor 162 as shown in fig. 9-19. In case of cross-correlation with the signal from the converter 51, the second valve wave in the double-valve wave will be eliminated by suitable means so that a cross-correlation can be made with the single valve wave from the converter 51.

Claims (5)

1. Method for telemetry of data in a borehole filled with drilling fluid, comprising generation of individual data-carrying fluid pressure signals down the borehole, which signals are arranged in a coded sequence in the fluid and this coded sequence represents the magnitude of a parameter down the borehole, generation of corresponding individual data-carrying pressure changes near the surface, which borehole fluid is also impressed by disturbing pressure changes caused by the drilling operations which thereby mask the said data-carrying pressure changes, where the pressure changes in the drilling fluid representing the superposition of the said data-carrying pressure changes and the said disturbing pressure changes are received near the surface, and the superimposed pressure changes are converted into electrical superimposed digital signals, characterized by that an electrical digital waveform signal is produced at the surface which represents the waveform of the aforementioned data-carrying pressure changes, that the electrical digital waveform signal is entered into the memory in a digital filter, that the digital filter is operated with the aforementioned electrical superimposed digital signals applied to the filter's input and the electrical digital waveform signals in the filter's memory, and that electrical signals are derived from the digital filter's output representing the aforementioned data-bearing pressure changes, where the masking effect of the aforementioned disturbing pressure changes is reduced to a minimum. 2. Method according to claim 1, characterized in that the digital filter performs a convolution of the electrical superimposed digital signals and the electrical digital waveform signals. 3. Method according to claim 1, characterized in that the digital filter performs a cross-correlation of the electrical superimposed digital signals with the electrical digital waveform signals. 4. Method according to claim 1, characterized in that a customized filter is used as digital filter. 5. Method according to claim 1, characterized in that a peak pulse filter is used as digital filter.