当前位置:首页 >> 电力/水利 >>

Torsional Resonance of Drill Collars with PDC Bits in Hard Rock00049204


A
SPE 49204 Torsional Resonance of Drill Collars with PDC Bits in Hard Rock
T.M. Warren, SPE, and J.H. Oster, Amoco E&P Technology Group

\ Scclety;f PetroleumEnglneefa\

Capyrfght 1938, Sc&ty

of Pefmleum Englneara, Inc.

This papw w prapare5 for presentation al the 1998 SPE Annual Technical Conference and EXhlbitlOn hekfln New Orleans, Louisiana, 27-30 September 1s2S. This paper was sekfsd for presentation by an SPE Prcgram Oommltfee following review of information contained In an abstract submitted by the author(s). Contenfa of the paper, as prasented, hava not bean reviewed by the %elefy of Patroleum Engineers and are subject to mrradon by the author(s). The material, as pres.entad, does not necassaffly raflsct any pasitlon of the Society of Patrolaum Engineers, M otfkers, or members. Papers presented at SPE maetings ara sub]act !0 publication revfew by Editoriaf Cemmitfeas of [ha SwIaty of Petrolaum Engineers, Elactronlc reprcductlon, dlsfribufion, or storage of any part of this paper fer mmmerclal purpoaas without the w?inen censent of tha SCclety of Petroleum Englneera Is pmhlblted. Parrnlsslon to reproduce In print Is restrfctad to an abstract of not mare than 300 words; illustrations may not be cnplad. Tha abstract must contain conspicuous acknowledgment of where and by whom the papar was prasented Write Librarian, SPE, P.O! sax S33836, FUchardson, TX 750S3-3SS6, U. S.A., fax 01-972-252-2435.

Abstract Formations encountered while drilling oil and gas wells are not homogeneous, This requires drill bits to be capable of drilling the most difficult rocks in the interval rather than the “average” rock and can result in a significant penetration rate penalty for cases where a few hard streaks preclude the use of PDC bits. Bit manufactures are actively trying to improve materials and bit designs to provide more robust bits, but in many cases it appears that some type of drillstring vibration limits the use of PDC bits in these hard streaks. Extensive data from drilling tests conducted at Amoco’s Catoosa test facility indicate that PDC bits are damaged very rapidly when they encounter certain hard reeks. The visual appearance of the damage suggests some type of vibration damage, but the normal cures for bit whirl and drillstring sticldslip do not seem to be very helpful. The analysis of nineteen wells where a downhole vibration sensor was run and PDC bits encountered severe damage indicates that torsional resonance of the drill collars could result in backward rotation of the bits. The observed vibration has been known to exist (in a theoretical sense), but is not recognized as a significant cause of PDC bit failure. The Catoosa drilling data shows conclusively that a wide range of PDC bits sustain this type of vibration when Laboratory tests demonstrate that they drill hard rocks. reverse rotation of PDC bits result in catastrophic cutter failure. The evidence collected so far suggests that reverse rotation due to torsional vibration is probably the cause of the rapid damage in certain hard rocks, but this postulation has not been proven because the sensor used in the tests was not adequately sensitive to say positively that the bit rotates backwards.

Introduction Often the formations encountered when drilling for oil and gas are not homogeneous and require a bit selection that is robust enough to drill the most dit%cult portions of the interval rather than one that is best suited to drill the “average” rock in the interval. This is especially pronounced in cases where most of the interval can be drilled effectively with PDC bits, but a few hard streaks preclude using them. In these cases a small extension to the range of rocks drilled by PDC bits would result in a significant increase in their application and allow their advantages to be utilized in zones that include hard streaks. The key to developing more robust bits is to better understand the fundamental mechanisms that cause their failure. Amoco has extensive field test data indicating that PDC bits are damaged very rapidly when they encounter certain hard rocks at our drilling site. The visual appearance of the cutters suggests they are damaged by vibration, but the normal cures for bit vibration do not seem to be very helpful in these harder rocks. Much of the publications on the effect of vibration on bit performance in hard rock concentrate on axial vibrations for roller cone bits and drill-string “stick/slip’’i-s or bit whirls”’1for PDC bits. The purpose of this paper is to discuss torsional resonance, another type of vibration that appears to be equally important in limiting PDC bit performance in hard rock. We will begin with a review of bit whirl and sticldslip to show that they are distinctly different, but the major objective of the paper is to show that torsional resonance of the drill collars is prevalent when drilling with PDC bits in hard rock and that the magnitude of the vibrations is also large enough to damage PDC cutters, This type of vibration is not new in the sense that is has not been reported before, but there are no known publications demonstrating that torsional resonance is likely to be a major con~ibutor to PDC bit damage. Driii-string Dynamics Sub The most effective method of detecting and identifying bit vibrations is to use a downhole instrument to measure the accelerations in the drill-string immediately above the bit. A downhole recording vibration data collection instrument (DDS12) designed and manufactured by a major service company has been run rather routinely since mid 1994 at Amoco’s Catoosa test site .27 Data observed from these runs have provided a much better understanding of the downhole

625

2

T.M. WARREN, J.H. OSTER

SPE 49204

“vibration environment. 13’26 The following discussion of torsional resonance relies heavily on measurements made with the DDS. The DDS tool records the accelerations from three orthogonally oriented accelerometers. The Z accelerometer responds only to accelerations directed along the longitudinal axis of the drill-string, while the X” and Y accelerometers respond to both lateral translations of the drill-string and to rotational accelerations around the longitudinal drill-string axis. It is somewhat convenient to think of the X and Y accelerations as composed of both lateral components in a Cartesian coordinate system and rotational components in a cylindrical coordinate system, as shown in Fig. 1. The DDS measurements are recorded as a time log of average accelerations and peak accelerations, and discrete files (burst files) of high data rate samples. Zanomi 12 provides more detail about the way the data is recorded. PDC 61t Whirl Backwards whirl is one of the predominant types of vibration that damages PISC bits, It is a condition where friction between the rotating bit and the rock causes the bit and lower portion of the drill-string to precess around the borehole in a direction opposite that of the drill-string rotation. Whirl is self sustaining because centrifugal force pushes the whirling bit into the borhehole wall or tapered portion of the bottom hole pattern, creating more friction. Thus after whirl is initiated it may continue as long as the bit rotation continues or until something interrupts bit rock contact. If whirl occurs without slippage between the contacting surfaces, the displacements, velocities, and accelerations can be determined exactly.c’7 The center of the whirling bit will make a circular path if it whirls in a circular hole, but all other points on the bit will move in noncircular paths. Bit whirl with bladed PDC bits is usually initiated by the bit face interacting with the bottom hole profile, rather than the borehole wall. (Whirl cannot occur unless the hole is at lest slightly larger than the bit.) The whirl radius will be near the center of the bit when the bit face initiates whirl in a hole that is nearly gauge, but moves outward as the lateral impacts enlarge the hole. In this case the whirl frequency is more dependent on the number of blades than the exact diameter of the hole. Figure 2 shows the displacements and accelerations for points located at the DDS sensor position (radius of 1.6”) and at the gauge (4.25”) when an 8-1/2” bit (4 blades) whirls in a five lobed bottom hole pattern and the hole has grown to the point where the whirl radius is equal to the bit radius (0.75” overgauge). The theoretical trajectories shown in Fig. 2 are in good agreement with cutter paths observed in bottom hole patterns obtained from laboratory tests. “14 The X and Y accelerations at each location have similar amplitude and the amplitude of the Y accelerations at the gauge of the bit are the same as on the inner part. The X accelerations have a small dc component due to the centripetal acceleration, thus they are slightly larger at the greater radius. Excluding this dc component, the X and Y accelerations are

nearly equal because whirl is primarily a harmonic lateral translation that is powered by the rotary motion of the drillstring. Bit whirl is routinely observed while testing PDC bits on lab drilling rigs, but many field observations, primarily bit damage in hard rock and downhole motor and MWD damage in soft rock, also confirm that whirl occurs and is just as violent as the lab observations indicate. Figure 3 shows a field case where the DDS recorded an incident of whirl that resulted in the destruction of the mud motor and a fishing job. A 12-1/4” PDC bit was drilling with low background accelerations until it encountered a depleted sand at a depth of 12960 ft. At the top of the sand, the peak lateral (X and Y) accelerations instantly jumped to about 200 G’s and the axial accelerations increased to about 50 G’s. After drilling about 40 ft, the motor housing parted and the lower part of the motor was left in the hole. The fish was recovered and the following damage was noted on the motor failure analysis report twisted off at the top of the stator comection, screw on stabilizer backed off, all housing comections over torqued, lower radial bearing split, lower stator comection cracked, offset housing connection cracked, rotor coupling “wallowed out”, and transmission shaft ball sockets “wallowed out”. This example shows the severity of the damage that can be caused by whirl, but in many cases the damage is less severe - such as only seeing the near-bit motor stabilizer backed off. At other times damage to MWD tools due to whirl are not even recognized as such. One of the burst files that was recorded by the DDS just before the failure occurred is shown in Figure 4. The lateral accelerations show classic evidence of whirl and support the conclusion that the failure was caused by bit whirl. The peak accelerations are well over those calculated from the expected bit motions, but when the raw signal is filtered at 25 hz to remove the higher frequencies, the accelerations are similar to the predictions. They have a peak magnitude of 5-7 G’s, but are more irregular than the differentiated motion equations would predict. This data is typical of that recorded during bit whirl and suggests that the actual motion experienced by a “whirling” bit is not a perfect whirling motion. This shows up in the raw acceleration data as very high acceleration spikes and in the filtered data as a discontinuity in the sinusoidal accelerations. Irregular bit motion during whirl is confirmed by the bit trajectory plots shown by Kenner.15 Bit Whirl Characteristics: q The lateral acceleration frequency is proportional to the rotary speed. q X and Y acceleration amplitudes are approximately equal, except for a dc offset that is not measured by the DDS. q X and Y acceleration amplitudes are nearly the same at any radial position on the whirling member. q The lateral accelerations are very noisy and contain high magnitude “shocks”. q The lateral accelerations and formation impact forces are proportional to the rotary speed squared. q Whirl generates high centripetal forces that cause whirl to

626

SPE 49204

TORSIONAL RESONANCE OF DRILL COLLARS WITH PDC BITS IN HARD ROCK

3

be regenerative and is eliminated only when the rotary speed is reduced to zero, even though the accelerations are reduced when the rotary speed is lowered. Drill-string Torsional Vibration Background In contrast to bit whirl which is primarily controlled by the bit and formation characteristics coupled with the operating conditions, another vibration dysfunction commonly referred to as sticldslip is more heavily related to drill-string characteristics. The bit is often still responsible for initiating sticldslip, but the phenomenon is primarily a torsional oscillation of the drill-string. In order to better describe drill-string sticldslip and to provide a foundation for explaining drill collar torsional resonance, a simple model (shown in Fig. 5) consisting of a thin tube (drill pipe) and thick tube (drill collars) is discussed. The model drill-string is suspended from a support at its top. This top support can be considered “fixed” if the pipe is rigidly attached to a large mass or it can be effectively “free” if it is supported on bearings. The vibration response is different depending on what condition exist at its ends. Closed form equationslc’17 can be used to estimate the natural frequencies for the system, but more accurate and versatile predictions can be made with a computer program such as that presented by Vandiver.18’19 This program (DSVIB 1) calculates accelerations, velocities, displacements, and internal forces in the drill-string for much more complicated geometries than can be calculated with closed form equations. The output from the program is typically shown as “mode The “transfer function” shapes” and “transfer functions”. shows the response at some point along the drill-stiing to a unit magnitude excitation of varying frequency applied at some other point along the drill-string. The mode shape is simply a plot of the maximum amplitude that is experienced at each point along the length of the drill-string when the entire drill-string is excited with a unit excitation at a particular frequency. If a unit torsional excitation (k) is applied to the lower end of the collars, torsional vibration can be observed at several different frequencies. At a very low, but unique frequency, the drill-string will oscillate as a torsional pendulum where the drill pipe is the spring and the collars are the mass. This type of vibration is generally referred to as “stick/slip” and in its extreme case the bit comes to a complete stop during each torque cycle. (Ir3 this paper we will refer to the torsional pendulum effect as “stick/slip” to distinguish it from torsional resonance of the collars, even though stictislip is sometimes used only to describe the case where the bit comes to a complete stop.) lUgure 6 shows the torque and displacement mode shapes for this excitation. The scale for each has been made dimensionless by dividing by the excitation torque, & The twist at the top of the collars is almost the same as it is at the bottom, ie, the collars behave as a fixed mass. The displacement of the tube is the greatest at the top of the collars and gradually decreases to zero at the surface. The internal torque results from the acceleration of the mass of the collars as they oscillate and thus the peak torque

increases linearly from a minimum value at the bottom of the collars to a maximum at the top of the collars. The peak torque is then nearly constant along the drill pipe at any point in time (neglecting friction). This torque is the internal torque experienced by each cross-sectional element in the tube, thus when a drill-string oscillates in this manner, all the drill pipe connections are exposed to the same oscillating torque as is observed at the surface. Of course there may be distributed friction along the string that will provide a steady component to the torque that increases from bottom to top. The distributed friction may also provide a damping to reduce the magnitude of the vibrations. A spring-mass system has no harmonics, so there is only one frequency where the drill-string will oscillate as a torsional pendulum as described above, but the system can vibrate in other ways. At a frequency considerably higher than the torsional pendulum natural frequency, the first natural frequency of drill collar resonance will be encountered. 20”21 Resonance simply means that a standing wave exists along the length of the collars and can occur for frequencies where the collar length is equal to 1/2, 1, 1-1/2, 2, etc, times the wavelength of the exciting frequency if both ends are either free or constrained. For a situation where only one end of the collars is constrained, they will resonate at frequencies where their length is 114, 3/4, 5/4, 7/4, etc times the wave length. Figure 7 shows the mode shape of the first natural frequency for torsional resonance of the collars in the model drill-string. Because the drill pipe is much less stiff than the collars, the collars are essentially free at their top (torsionally speaking) and can resonate as a prismatic bar suspended on bearings. The bottom also behaves as a free boundary based on the evaluation of vibration data collected with the DDS (even when drilling with a PDC bit). For this case the maximum torque occurs near the mid point between the ends of the collars and is much higher than any torque observed at the surface. Drill-String Stick/Slip. In field drilling operations the drillstring often behaves as a torsional pendulum as discussed above for the idealized drill-string composed of drill pipe and drill collars.3’4’5 Drill-string stick/slip is quite easy to recognize in the field because the torque oscillates with a rather uniform period of several seconds and the torque oscillation is often accompanied by a corresponding oscillation in rotary speed, Fearzz shows field examples where the oscillations are developed to different degrees, ranging from mild to severe where sticldslip causes the bit to stop during each cycle. For these cases the bit may come to a full stop and then race ahead as shown by the downhole bit speed data recorded by Jansen23 and shown in Fig. 8. Large fluctuations in the torque (Fig. 9) can be observed at the surface due to the acceleration of the collars when they undergo these wide changes in rotation speed. Often the driller may attribute the torque fluctuations to the bit “taking a bite” when in fact they are simply the result of the torsional pendulum effect. Damage from drill-string sticldslip primarily falls into two

627

4

T.M. WARREN, J.H. OSTER

SPE 49204

categories: drill-string damage and bit damage. Drill-sking connection damage can result from the high cyclical torque, particularly when the torque peaks exceed the make up torque. Recall from the earlier discussion that every drill pipe connection experiences a torque equal to the static torque (less filction) at that point plus the dynamic torque component observed at the surface. This is particularly critical when running a tapered string. During periods of sticldslip, the instantaneous bit speeds are much faster than the rotational speed observed at the surface as shown in ~lg. 8. These periods of rapidly accelerating bit speed cart cause short duration episodes of whirl where high lateral impacts can damage PDC cutters. This damage can be particularly bad when the stick/slip is fully developed where the bit completely stops rotating for a significant portion of the cycle. The bit then accelerates to a speed much higher than the average rotary speed which provides ideal conditions for the bit to whirl for a short duration of each sticldslip cycle. Intermittent whirl similar to This this has been observed in laboratory drilling. phenomenon is consistent with Fear’s22 observations and his conclusion that anti-whirl bits tend to help prevent the damage from stick/slip. Drill-string sticldslip is initiated by friction somewhere rdong the lower end of the drill-string. In some cases this is simpIy borehole wall fiction and may be distinguished by the fact that the drill-string oscillates even when the bit is off bottom: In other cases the friction comes from the bit face and gauge contact. It is often a sign that the gauge of the bit is worn and is used as a bit wear diagnostic tool, particularly for roller cone bits. Stick/slip is more prevalent with PDC bits in hard rock than it is with rollercone bits. It appears to be worse with PDC! bits because there is often a negative correlation between bit rotational speed and bit torque? As the bit speeds up, the torque decreases and provides a feedback that initiates and sustains the sticldslip torsional oscillation of the bit. Stick/slip occurs for a range of rotary speeds and WOB, with the range being determined by the borehole, formation, drill-string, and bit characteristics. Generally, for any of these conditions though, the sticldslip can be reduced by increasing the rotary speed and reducing the WOB, but the smooth region may be outside the parameters that provide an acceptable penetration rate or outside the operating limits provided by the rig. The stick/slip can also be mitigated by using a special controller (one version is called “soft torque”) on the rotary drive that varies the energy provided to the rotary drive unit to interru t the feedback that allows the oscillations to build up 24,2 Figure 9 shows an example of drill-string stick/slip recorded at Catoosa for a drill-string consisting of 597 ft of 61/8” chill collars and 842 ft of 4-1/2” aluminum drill pipe. The frequency of this torsional vibration is 0.7 hz. DSVIB 1 predicts the frequency for drill-string sticldslip for the conditions shown in Fig. 9 to be 0.56 hz for a completely stiff top boundary condition, but the model can be tuned to match the observed frequency by inserting a less stiff element at the

top of the drill-string. This frequency is higher than normally experienced in the field because of the shallow depth of the Catoosa wells. Drill-string sticldslip is easy to identify from surface observations, but is often not discernible from the DDS tool burst files because of its low frequency and the short DDS sample time. Stick/slip may be indicated by the peak X accelerations being higher than the peak Y accelerations, but it is doubtful that sticklslip could be identified from the DDS data if any other form of vibration were occurring at the same time. This would be even more true for field cases where the period of the oscillations are longer. Drill-Strins? Stick/S1i~ Characteristics: Controlled by dr~l pipe, depth, collar mass, friction, and bit properties. Frequency is nearly independent of rotary speed. Period is quite long (greater than 1 second). Has a higher X than Y, but the DDS tool is rarely programmed with a sample interval long enough to identify it from the burst files, Maximum instantaneous bit rpm is several times the average surface rpm. May cause short duration episodes of bit whirl each time the bit accelerates. May be caused by either friction along the drill-string or by bit friction. Obvious evidence of torque cycling that can be observed on the rig floor. Torque oscillation observed at the surface is applied to each drill pipe connection. Can usually be eliminated by increasing rotary speed or empIoying “a“soft torque”. Drilling Test Data There have been many PDC bits of various sizes, designs, and brands run at Catoosa over the last few years. These bits include most of the concepts that are currently incorporated in commercial bit designs aimed at drilling rocks on the upper limit of hardness that can be drilled by PDC bits, including anti-whirl designs, At Catoosa it is rare for one of these bits to drill completely through the Mississippi limestone section without encountering catastrophic damage to the diamond cutters. A zone of very hard rock, dubbed “the wall”, is encountered at a depth of about 1376 ft that almost always results in severely damaged PDC cutters. (Unconfined triaxial tests of cores recovered from this interval show compressive strengths of 40,000 - 50,000 psi). Historically we have believed that this damage was caused by whirl, even when anti-whirl bits were run. It was assumed that the bit cutters could not penetrate the rock adequately to maintain the lateral force directed to the low friction pad to prevent whirl. At least nineteen wells have been drilled through this zone with PDC bits while running the DDS. Since the initial evaluation of the data from these individual wells was based primarily on peak and average accelerations, the collective data was re-examine to see if the burst files shed any

628

SPE 49204

TORSIONAL RESONANCE OF DRILL COLLARS WITH PDC BITS IN HARD ROCK

5

additioml light on the pervasive cutter damage. Figure 10 shows driIling data from two runs in one of the wells that is typical of the results of drilling into the “wall”, The bit was examined at 1254 ft and observed to be like new, but when examined again at a depth of 1404 ft it had severely damaged cutters as shown in Flg 11. A constant rotary speed of 120 rpm was run until the wall was encountered and was then increased to try to improve the penetration rate. (In other runs the rotary speed was kept constant and/or reduced and the cutter damage was similar.) Previous work has shown that a large wear flat rapidly develops on the carbide after catastrophic failure of the diamond layer.6 The peak and average DDS data (shown in Fig. 10) indicate that the lateral accelerations increased at the top of the Mississippi limestone and became quite high for the last twenty feet of chilling. The average Y accelerations were much higher than the average X and Z accelerations, but the peak X and Y accelerations were also high. DDS burst data was recorded at points indicated as A -H. Some of these points (A - D) had quite low vibrations because no point in the run encountered vibrations that would have overwritten the low amplitude recordings. For the second run the vibrations were much higher and burst data was stored for only the highest vibrations because the tool records only a fixed number of burst files and higher vibrations overwrite lower vibrations. DDS burst data was recorded at points indicated as A - H. Some of these points (A - C) had quite low vibrations because no point in the run encountered vibrations that would have over written the low amplitude recordings. Figure 12 shows the data recorded at point C. The vibrations are quite low and rather random, The power spectrum shows that there is no power in either the X or Z accelerations, but there is a very low power level in the Y accelerations from near zero to about 40 hz. Certainly there is nothing recorded in this data that should be damaging to the bit, but the data shows broad band, low amplitude noise that can potentially excite the natural frequencies of the drillstring. Figure 13 shows acceleration data with a very strong harmonic Y acceleration at a frequency of 20.5 hz recorded at point D while drilling through a very thin hard streak, Neither the X or Z accelerations are significant, but there is considerable energy associated with the Y accelerations. Similar vibrations (but higher amplitude) were recorded at points E - H and the frequency of the vibration did not change as the rotary speed was increased from 116 to 196 rpm for these points. Since the X and Y accelerations are so different in magnitude and the frequency is not determined by the rotary speed, the vibrations are clearly not associated with whirl. Their high frequency indicates they were not caused by drill-string sticldslip. The most logical source of these accelerations is drill collar torsional resonance. As discussed later, the DDS data did not have sufficient resolution to simply integrate the accelerations to determine the collar motion, In order to indirectly confii the source of the vibrations, the expected DDS response to various rotational speed variations was

calculated and compared to the measurements. Secondly the DSVIB 1 program was used to determine the torsional resonance frequencies for various assemblies that were run for the Catoosa tests. The DDS burst files indicate that a sinusoidal Y acceleration with a magnitude of 10 G’s and a frequency of 20hz is fairly representative of the harmonic vibration data observed for points E - H. If torsional resonance of the drill collars is the source of these accelerations, then a fairly large sinusoidal variation of the collar rotational speed would be expected. The response of the X and Y accelerometers in the DDS were calculated for an assumed range of speed variations. A good match between the calculated and observed accelerations was obtained by varying the rotational speed from -60 rpm to 300 rpm. These values were chosen to simultaneously match the mean drill-string rpm (120 rpm) and the peak Y acceleration of 10 G’s. Obviously the conditions chosen (note negative rpm) have to provide an a}’erage rotation speed given by the average surface rotational speed, otherwise the twist in the drill-string would continuously accumulate, Other than the shift in X accelerations due to ac coupling, the similarity of the calculation to the observation were so close as to be quite convincing that the cause of the vibration is correctly understood. The natural frequency of the collars for torsional resonance can be estimated with the equations for a cylindrical beam and calculated more exactly for a particular geometry with the DVIB 1 program, as discussed above. Figure 14 shows the frequencies expected for the first five modes calculated from the closed form equation, In general, frequencies of 8 to 52 hz can be expected for typical collar lengths. Many cases of harmonic vibration that correlated with the natural frequency of the drill collars as calculated from DSVIB 1 were recorded The primary frequency was rarely at the by the DDS, fundamental mode, but was at one of the higher modes. Figure 15 shows a comparison of the DSVIB 1 predictions to a case where there were multip!e harmonics of about the same amplitude recorded with the DDS. The most interesting thing to notice is that the frequencies match well, but that there are no odd numbered modes. Based on the above evidence it was concluded that the large amplitude harmonic accelerations observed while drilling hard rock with PDC bits was caused by torsional resonance of the drill collars. At this point the focus shifted to understanding more about the impact of torsional resonance relative to other types of bit vibration. One major difference between whirl and the torsional collar resonance is that whirl is primarily a lateral translation of the whirling member while the torsional resonance is .a rotation about the member’s axis. The significance of this is that the accelerations recorded by the DDS tool for whirl are representative of all points on the bit, but the motions of individual cutters can be in almost any direction. For torsional resonance, the linear accelerations increase with the square of radial position but the motions are circumferential. Since the DDS sensor is near the center of the motion, the linear accelerations at the OD of the bit will be much higher than measured by the DDS. The accelerations at

629

6

T.M. WARREN, J.H. OSTER

SPE 49204

the gauge of an 8-1/2” bit are about 7 times higher than measured by the DDS, while the gauge accelerations of a 171/2” bit are 30 times higher. If the DDS records amplitudes of 10 G’s then the actual gage acceleration will be 70 and 300 G’s. Effect of Resonance on Bit Speed and Reverse Rotation. Since reverse rotation was required to match the accelerations observed with the DDS, two methods were put to use to determine if reverse rotation actually did occur. First, the Y accelerations were assumed to be tangential accelerations with no lateral translational component and were integrated to determine tangential velocities. The integration introduces a constant of integration that can be determined from the average rotary speed if it is assumed that the acceleration has only rotational components. For the DDS data, it was found that the data often included a small acceleration offset that caused the velocities to be either increasing or decreasing with Clearly this was not the case, so an offset was time. determined that forced the time-averaged velocity to be constant. (The simplifying assumptions were required because of the instability of the sensors around zero and the associated ac coupling). The second approach used to determine if the velocities were negative was to identify the component frequencies from the power spectral density and then run the recorded data through digital notch filters to extract the individual component magnitudes. A composite acceleration consisting of exact sine terms was then determined using the frequency and amplitude of the observed components. This function could then be exactly integrated to determine the velocity. Figure 16 shows the results of applying both of these processes to the data recorded at point D on Fig. 10. Both methods indicate that the instantaneous rotational speed varies from near zero to nearly 300 rpm. Both methods also show that there is a possibility that the cutter moves backwards, but the data is insufficient to conclude this with any confidence. Figures 17-19 show the accelerations recorded for points E G, the power spectrum of the vibrations, and the bit velocity determined from integration. These points show more substantial evidence that the cutters rotate backwards. Similar results were seen with the superposition technique and for many other cases recorded with other bits, but a definite conclusion of reverse rotation will require measurements with a more accurate sensor. Once it was established that the vibrations observed on the DDS were caused by resonance of the drill collars and there was a high probability that this could cause the cutters to move backward, attempts were made to evaluate the significance of the vibrations and to determine other characteristics of them that might lead to a reduction in their potentially detrimental affects. Whether or not a cutter moves backwards depends on the amplitude of the accelerations, the frequency of the accelerations and the average rotary speed. Figure 20 shows the amplitude/frequency regions for 60 rpm and 120 rpm where backwards rotation can occur for a simple harmonic motion. For a typical frequency of 20 hz, any accelerations

over 3.5 G for 60 rpm and over 6.5 G (at the sensor position) for 120 rpm resuks in reverse rotation. This means that the cutters travel backwards once each cycle, ie 20 times per second, The amplitude of DDS accelerations observed indicates that a gauge cutter on an 8-1/2” bit may travel backwards as much as M“ each cycle. Figure 21 shows a PDC cutter schematic with force directions for both forward and reverse rotation. The forces acting on the diamond layer for forward rotation keeps the diamond in compression where it is strongest. When the rotation reverses, the diamond layer is placed in tension where it is very weak. This can cause a large part of the diamond to flake off as indicated by the “crack” in the left side of Fig. 21. Lab tests were run with a PDC bit manufactured with premium cutters to observe the effect of reverse rotation on cutter failure. A pilot hole was drilled in Carthage limestone and the bit was rotated backward with various WOB ‘s. The amount of rotation was determined by the minimum increment that could be applied by the rig and was typically less than 20°. At the fust weight of 1,000 lb, one cutter failed as shown in Fig. 22. Tests were continued at weights up to 10,000 lb without any further cutter failure. Another series of tests were run with a much stronger sandstone. In these tests cutters failed at 1,000 lb and 3,000 lb. Because of the extensive failure seen at 3,000 lb, no higher weights were run. Examples are shown in Fig. 22 for the failures that occurred for these tests. In general, large areas of spalling occurred near the point of contact with the formation, This occurred even though the cutters were “premium” cutters where the diamond carbide interface is non-planar to reduce stresses and make a stronger cutter. Some of the failure surfaces looked very similar to the failures seen at the “wall” at Catoosa, while others appeared more severe, Clearly the reverse rotation caused severe cutter damage in the hard sandstone, but unfortunately the conclusions from the test are somewhat compromised because two cutters also failed while drilling the pilot hole, It is possible that the axial contact force between the cutter and rock was high enough to create tensile stresses in the cutter, w’hich caused it to spa]]. We can not say conclusively at this time that similar spalling could not occur from forward rotation of the PDC bit in rock like the sandstone used for this test. When the cutter photographs shown in Fig. 11 are examined in light of the calculated bit speeds and observations about negative rotation on cutter chipping (Fig. 22) it is easy to conclude that there is a very high likelihood that the cutters were damaged by reverse rotation in hard rock. It is also evident that some amount of reverse rotation can be tolerated without failure for some rocks and cutters. Prevalence of Collar Resonance. ISDS data was available for a total of 19 wells drilled with a variety of PDC bits ranging in size from 7-7/8” to 12-1/4”. These bits included samples of almost all the styles that various manufacturers use for hard rock applications. In every case, the data indicated that the collars oscillated at a resonant frequency and the bits

630

SPE 49204

TORSIONAL RESONANCE OF DRILL COLUARS WITH PDC BITS IN HARD ROCK

7

experienced a very unsteady rotating speed while drilling hard rock. The torsional frequencies for different collar lengths shown on Fig. 23 were compiled from the Catoosa data and are compared to the first two modes of vibration calculated from the beam equations, In general the frequencies most often seem to correlate with the second mode and have a slightly lower value than the closed form solution predicts. In only a few cases were examples of the fundamental observed, but many cases of higher modes that are not shown were observed. The collars can also resonate at frequencies that are quite high. Both DSVIB 1 predictions and DDS measurements show evidence that the collars may resonate at frequencies ranging from 150-400 hz, We do not think these high frequencies are damaging, but high frequencies were clearly seen on the DDS data. The following points summarize the observations from the 19 wells that were examined: q Torsional resonance of the collars was observed in every case where hard rock at “the wall” was drilled with a PDC bit. q Torsional resonance was not observed while drilling soft rock. q Resonance with bit sizes ranging from 7-7/8” to 12-1/4” was observed. s No difference between the torsional oscillations caused by anti-whirl and standard PDC bits was evident. q The instantaneous bit speed can be several times the average bit speed. q There is a high probability that the bit rotates backward during resonance. q The frecwencies observed generally agree with the expected-natural frequencies ;f the dr;ll c&rs. Other Tests Other specific tests were conducted to better understand the torsional resonance problem, to confirm that it was not just a “Catoosa” effect, and to evaluate possible solutions. Aluminum vs. Steel Pipe. The standard drillpipe used at Catoosa is 4-1/2” aluminum pipe with steel tool joints. One of the first questions explored was whether the aluminum drillstring made the torsional resonance problem worse at Catoosa than it might be when drilling with steel drillpipe. Figure 24 shows a comparison of the resonant frequencies for 20 collars with 4-1/2” aluminum drill pipe and 4-1/2” steel drill pipe, For all practical purposes the drill pipe made little difference in the calculated frequency response of the collars. Even though the calculations indicated that there should be little difference between the aluminum and steel pipe, a test was conducted to conclusively put the issue to rest. Figure 25 shows a portion of a test that included a run where the aluminum drillpipe was replaced with steel drillpipe after drilling into the “wall”, The steel pipe was run for about fifty feet and then replaced with the aluminum pipe again. The average Y accelerations with the steel pipe were just as high

(possibly higher) as with the aluminum pipe. The frequencies observed were slightly higher than with the aluminum pipe (26 hz instead of 17 hz), indicating that the energy was concentrated at the third mode instead of the second. The conclusions from these tests were: 1) collar resonance occurs with steel drillpipe just as it does with aluminum drillpipe, 2) the resonance frequencies seem to be shifted higher when steel drillpipe is used compared to aluminum drillpipe, and 3) the average resonance amplitude may be higher for steel drillpipe than for aluminum drillpipe. Rotary Speed and WOB Effects, Several tests were run to evaluate the effect of rotary speed on torsional resonance of the drill collars. The collar resonance is similar in many ways to the drill-string stick/slip type of vibration that can be eliminated by rotating the drill-string at speeds generally greater than 90 rpm. Because of the similarity between drillstring sticldslip and drill collar resonance, a stability region for collar resonance was expected. Figure 26 shows WOB variations ranging from 10 to 30 thousand pounds at rotary speeds of 60, 120, and 180 rpm. All the tests at 120 and 180 rpm had torsional resonance exclusively at a frequency of 17,5 hz. The amplitude of the accelerations increased with rotary speed, but generally remained constant as the V+70Bincreased. One exception was the low WOB, high RPM case where the amplitude was significantly less. It is possible this indicates that there is a threshold WOB and rpm combination where the torsional oscillations do not occur, but if so, it is probably not relevant to hard rock drilling. The frequency of the vibrations at the lowest rotary speed included several of the higher modes rather than being at a single frequency. Other examples, including the steel drill pipe, showed the same trends. There also appeared to be a slight trend for sharp bits to have a higher tendency to excite lower modes than worn bits. This would indicate that sharper bits might be more prone to damage from torsional resonance than dull bits. There were no observations in any of the data to suggest that there were particular combinations of rotary speed and bit blade count that made the torsion resonance worse. The general observations from changing the operating conditions were: q Acceleration amplitude increases with rpm. q Acceleration amplitude increases very little as WOB increases. q The torsional resonance at low rpm tend to be at the higher harmonics. q There maybe a WOB and rpm stability zone where the oscillations do not occur, but it is outside the normal operating range for hard rocks. q The probability of the bit moving backwards is reduced at low rpm because of the tendency to excite higher harmonic modes. Roller Cone Bits. Drill collar torsional resonance was observed to occur in a few cases when drilling with roller cone

631

8

T.M. WARREN, J.H. OSTER

SPE 49204

bits. It occurred much less frequently and the amplitude was much lower than with PDC bits. Can Drill-String Stick/Slip and Collar Resonance Occur Together? Several incidence were recorded while drilling hard rock with a PDC bit where the transition to and from drill-string sticldslip was recorded similar to that shown in Fig. 9. In each case the torsional resonance died out immediately after the sticldslip started and started up again as soon as the operating conditions were changed enough to eliminate the sticldslip. It seems that these two modes of vibration may be mutually exclusive. There is some logic to suggest this because the collar resonance would substantially change the bithck torque interaction that is thought to be a primary feedback mechanism to sustain the sticldslip vibration. Transition ColIars. To some extent the torsional vibratory response is affected by the impedance match between the top of the collars and the bottom of the drill pipe. The potential for reducing the resonance by placing transition collars (where the stiffness is gradually reduced from that of the collars to that of the drill pipe) between the collars and pipe was evaluated. Cases were run with DSVIB 1 where five transition collars were used and their length varied from 30 ft each to 7.5 ft each. The result was that the frequencies could be varied, but the resonance was not eliminated, Torsional Shock Sub. Cases were also n.rn with DSVIB 1 to evaluate the potential to reduce the torsional resonance by placing a torsional shock sub between the bit and collars. The shock sub was placed immediately above the bit and was simulated as a one-meter long element with much lower stiffness than the collars and a high degree of damping. Choosing the appropriate parameters for stiffness and damping eliminated the low frequency resonance, but a fairly wide band of higher frequency resonance remained. The potential for the bit to rotate backwards is greatly reduced if the resonance is forced to the higher frequencies, so it is quite possible that a shock sub with adequate properties could be designed that would be useful. Could Torsional Resonance Be Useful? When drilling with PDC bits on the laboratory drilling rig torsional vibration is often observed, particular] y with duller bits and high WOB. (It is actually stick/slip, but because of the drill-string geometry it occurs at a frequency of about 60 hz.) When this vibration occurs, the penetration rate often increases significantly. No similar increase in ROP has been seen in the Catoosa tests in hard rock, but no torsional oscillations were observed in softer rock so that a comparison could be made. This leaves open the question of the potential to increase ROP by inducing torsional resonance in softer rocks. Summary Itis quite evident that torsional resonance of the drill collars can result in large amplitude harmonic vibrations when

drilling with a PDC bit. Drill collar resonance was often observed when PDC bits drilled hard rock, but never in soft rock. Bit design did not seem to affect the occurrence of collar resonance, but possibly, lower frequencies are excited by sharper bits. Low rotary speeds are likely to excite higher harmonic frequencies which are less damaging. The amplitude of accelerations due to collar resonance increases as the rotary speed increases, but increases to a lesser degree as the WOB increases. There probably is a rpm/WOB region where collar resonance does not occur, but it is outside the normal PDC bit operating range for hard rock. Collar resonance is not detectable from normally observed surface parameters, but may be observed on the spectrum analysis of axial vibrations at Catoosa depths. collar resonance and sticldslip appeared to be mutually exclusive in the limited data available at the initiation and termination of sticldslip, When collar resonance occurs, the bit speed varies widely at frequencies from 8 to 52 hz, Collar resonance probably causes the bit to rotate backwards. The initial cutter damage at the “wall” at Catoosa is consistent with negative bit rotation and it appears altogether likely that PDC bit cutters can be damaged in hard rock by reverse rotation caused by torsional resonance of the drill collars. References
1, Drrfeyete, M. P., and lfeMeUSe, H.: ‘Detectingand Monitoring of dre Stick-Slip Motion: Field Experiments,” paper SPE 21945 presented at the 1991 SPE/JADC Drillig Conference, Amsterdam, March 11-14. Dawson, R., Lin, Y. Q., and Spanos, P. D., “Drillstnng Stick-Slip Oscillations”, Presented at the SEM Spring Meeting (19S7). Halsey, G.W. et aL: “Drill-string Torsional vibrations Comparison Between Theory and Ekpentnent on a Full-Scale Research Drilling Rig,” paper SPE 15564 presented at the 1986 SPE Annual Technical Conference and Exhibition, New Orleans, Oct. 5-8. Kyllingstad, A. et al,: “A Study of Slip/Stick Motion at the Bit;’ SPEDE (Dec. 1988) 369-73. Brett, J.F.: “The Genesis of Bit-frrduced Torsional Drill-string Vibrations: paper SPE/IADC 21943 presented at the 1991 Drilling Conference, Amsterdam, March 11-14. Brett, J.F., Warren, T. M., and Behr, S.M.: “Bit Whirl - A New ‘Ilmry of PDC Bit Faihsr%” paper SPE 19571 presented at the 1989 SPE Annual Technical Conference and Exhibition, San Antonio, Oct. 8-11. Langeveld, C.J.: “PDC Bit Dynarnicsv paper JADUSPE 23867 presented at the 1992 Drilling Conference, New Orleans, Febuary 18-21. Warren, T.M., Brett, J.F., and Sinor, L.A.: “Development of a WhirlResistant Bit7 paper SPE 19572 presented at the 1989 SPE Annual Technical Conference and Exhibition, San Antonio, Oct. 8-11. Sinor, L.A. and Illerhous, R.: “Anti-Whirl PDC 8its - Evolution and Field Experiences; Oil Gas - Eurorrean Magazine (March 1993), 29-35. Weaver, G.E. and Clayton, R. I., ‘“A New PDC Cutting Structure Improves Bit Stabilization and Extends Application into Harder Rock Types,” paper SPFJL4DC 25734 presented at the 1993 Drilling Conference, Amsterdam, Feb. 23-25, 1993. Clegg, J. M., “An Analysis of the Field Performance of Anti-Whirl PDC Bits;’ SPIWADC 23S68 presented at the 1992 IADC/SPE Drilling Conference, New Orleans, LA, February 18-21. Zarmoni, S.A., Cbeadram, C.A., Chen, D.C.K. and Golla, C.A.: “Development and Fleld Testing of a New Downhole MWD Drill-string Dynamics Sensorp paper SPE 26341 presented at the 1993 Annual Teclrnkaf Conference and Exhibition, Houston, Getober 6-9, Dykstra, M. W., Chen, D. C. K., Wasren, T. M.; and Zannoni, S. A.: “Expenmerttal Evaluations of Drill Bit and Drill String Dynamics; paper SPE 28323 presented at the 1994 SPE Annual Technical Conference and Exhibition, New Orleans, Sept. 2S-28.

2.

3.

4. 5.

6.

7.

8.

9. 10.

11.

12.

13.

632

SF% 49204

TORSIONAL RESONANCE OF DRILL COLLARS WITH PDC BITS IN HARD ROCK

9

14.

15.

16.

17. 18. 19. 20.

21. 22.

23.

24.

25.

26.

27.

Behr, S. M., Warren, T. M., Sinor, L.A and Brett, J.F.: “Three Dlmerrsional Modeling of PDC Bits; paper SPE 21928 presented at the 1991 SPWIADC Drilliig Conference, Amsterdam, Mamh 11-14. Kemrer, J. V. and Isbell, M. R.: “Dynamic Analysis Reveals Stability of Roller Cone Rock Bits’’, paper SPE28314pmsented atfhe 1994 Amma1 Tectilcal Conference and Exhibition, New Orleans, LA, September 2528. Timoshenko, S, Young, D. H., and Weaver,Jr, W Vlbrat ion Problems in . . ~ fourth addition, John Wiley & Sona Inc., New York Ctiy (1974). ‘ ical Vibrations, Dover Publications, Irtc, New brt Hertog, J. P., _ York City (1985). . . 1 u ser’s Manu&, Proeram. A Drdls!au Vibration mv~l. Massachusetts Institute of Technology, 1991. H. Y. Lce~ Axial Vibration and Wave Prooaza tion in ~Ph.D. thesis, M. 1. T., 1991. Clayer, F., Vandiver, J. K., and Lee, H. Y.: “The Effect of Surface and Downhole Boundary Conditions on the Vibration of DrilIstringsY paper 20447 presented at the 1990 Annual Technical Conference and Exhibkion, New Orleans, Sept. 23-26. Dareing, D, W., ‘Drill Collar Length is a Major Factor in Vibration Control”, Jour. Pet. Tech , (April 1984), 637-644. Fear, M.J., Abbassian, F., and Ptiltt, S.H.L.: “The Destruction of PDC Bits by Severe Slip-Stick Vibration: paper SPWL4DC 37639 presented at the 1997 SPIYJADC Drilling Conference, Amsterdam, Mamh 4-6. Jansen, J.D.,van der Steen, L, and Zachariassen, E.: “Active Damping of Torsional DrNstrirrg Vibrations with a Hydraulic TopdriveT paper SPE 28911 presented at the 1994 European Petroleum Conference, London, October 25-27. Hrdsey, G.W., Kyllingstad, A., and Kylling, A.: “Torque Feedback used to Cure Slip/Stick Motion} paper 18049 presented at the 1988 Amrual Technical Conference and Exhibition, Houston, Ott.2-5. Sarranikone, P., Kamoshima, O., and White, D. B.: “A Field Method of Controlling Drillstring Torsional Vibrations;’ paper IADC/SPE 23891 presented at the 1992 Drilling Conference, New Orleans, February 1821. Warren, T. M., Oster, J.H., Sirror, LA,, Chen, D.C.K.:“Shock Sub Performance Tests: paper SPE 39323 presented at the 1998 IADC/SPE Dnlliig Conference, Dallas, March 3-6. Behr, S.M., Oster, J.H., Warren, T, M., “Amoco”s Test Facility Develops New Drilling Technology”, World Oil, (October 1995), 37-43.

Time, sec X Position, in. Motion and Acceleration at Sensors Position

0.4

X Position, in.
Fig. 2- Motions

Time, sec Motion and Acceleration at Gauge of Bit

and sccelerstlons for bit whirl (150 rpm).

‘“m
Imo ism lSSYI

i’=
tan ma molo

Wsr

x

Fig. 3- Field recorded accelerations durfng an episode of bit wM.

. $ 11m &sra]
Wg G%

‘-FimG1
xAd$ y~

<Ma

[

2-1/4”

k----

omaaos~ouom

1
ikda
1Axh

Whirl initiated at / depleted sand

H’

ZAd9

Motor Failure

Tti?kd off at t~ of statw onrndon . All housing oxvradlom toqud

. Screw on atatilizw barked off ovw

. Lower radlll beariIvJ @ii . Low atatm connection CrecAed . offset IKwsicg mnntion . Rotcf qiing

.1bhd

2sSii?

“wallowed our-

. Trammksion .4xM ball ac.dwts “vdlcwtw our

DC k .

mwm*snt6,1m=

*

Y

,

‘.

tool L Translations center
Note: ac coupled accelerometers may cause data offsets.

.,54 0

1 03 02 0.s

1 0.4 Q5

i as

Time, saconds

Fig. 1- DDS accelerometer orlentetlona.

~o””””””’””” ~ I --::_I ,
,o=m-10w70 ww, mooco 0.To150~ofio3,~,~ mm, FI’WWICY, hz

. . . ............ . ............ ... ..........

OA50~

Mmnlh

Fig. 4-

Burst date recorded during the high vibrations.

633

10

T.M. WARREN, J.H. OSTER

SPE 49204

+-EEl
‘a 200 &
_t

Downhole \ Surface

46.25”

I

o 0
10 20 30 40 50 after Jansen, et al

Time (seconds)
FicI. 5- Geometry used for example example drillatrlng co-mponents. Fig. 8- Field example of downhole bit apeed during aticldallp.

o E
100

g200 1g 300 k E

?/400 # 500 n 600

,oo~
0.00

E
0.05 0.10 0.15

Wd
!– 8 i-

1000
-1000 I

I

1

I

I

o

25 Torque

50

75

50

100

Ang. Displ.

150 Time, (seconds)

200

250

Fig. 6- Mode shapes for toralonal pendulum (atlcldsllp).

Fig. 9- Drillstring aticklalip at Catooes (1436 f$ 120 rpm, 1.4 sac period).

’50 m

EiziG

Q (ft-lbs]

,, :,,

o

Iw

200 300 400

0.00
Ang. Displ.

0.01
Ill

Torque

[
0 Wmomolc+o

\

Fig. 7- Mode shspes for oollars and drillplps at the first natural frequency of collar resonsnce.

10 20

Fig. 10- Typical Catoose well where PDC bit failed at the “wall”.

634

SPE 49204

TORSIONAL RESONANCE OF DRILL COLLARS WITH PDC BITS IN HARD ROCK

11

Natural
60
N 50 .c

Frequencies

for Free/Free

Boundaries

s

g 40

~30 u.

Fig. 11- Typical cutter damage at the “wall”.

0 0
200 400 600 800 i 000

Drill Collar Length, 10 .5 Go -5

ft

Fig. 14- Natural frequencies of typical collar lengths.

-10 ~
o 0.11 0.1

0.2

0.4

0.5

Time,s%
0
20 40 60 60 100

Frequency, hz
0
Fig. 12-Typical at point C.

20

Fr%quencyYhz

so

100

accelerations for smooth drilllng with PDC bit

10

100 60 80 40 Frequency, hz Fig. 15- comparison of measured and observed frequencies of collar

o

20

5 Go -5 -lo 0
0.1

resonance.

Speed

from

Integration

of unfiltered

data,

(1 16 rpm)

0.2

Time, s~~

0.4

0.5

o

0.1

0.2

0.4

0.5

Time,
o

s%

Frequ%cy, hz

100

0

0.25

0.5

Time, sec

Fig. 13- typlcsl DDS accelerations recorded during drill collar resonance at point D.

o

0.02

0.04

0.06

0.08

0.1

Time, 635

sec

Fig. 16- bit apeed estimated at point D.

12

T.M. WARREN, J.H. OSTER

SPE 49204

20 10 GO -lo -m

o

0.1

0.2 Time, &3

0.4

0.5

RPMfrom Imemation

o
Fig. 17- Acceleratlona recordad at point E (117 rpm).

10

20

30

40

50

Frequency, hz
Fig. 20- Conditions corresponding to reverse rotation (elmpla sinusoid without hsrmonlcs).

20

10 GO -lo
-20

+ ~~
., .,.,::: ..> ,,.> *j~’.

e
-... ----.. --.S--.,” -----. .

Diamond Wr in compreswon

0
20

0.1

0.2

0.4

0.5
.“-. ..?---..- .-... ..--””-

Time, s~~
RPM from Integration ‘1 A A A /1 /1 J1 /\

lll\lll\l \l
v

A

-------

“-

@

u

u

v

Forwa rd Rotation
Fig. 21- Reverse rotation causes tensile stresses in diamond lsyer.

o

100
FKr4Ju%Cy, hz

“--

o

0.1
Time, eae

0.2

Fig. 18- Accelerations recorded at point F (145 rpm).

20 10 GO -lo -20

0

0.1

0.2
Time, s~~

0.4 RPM from
=% %

0.5
Integration

f;= o
Frequ~ncy, hz 100

g

200 ioo -10:

0

0.25 Time,sec

0.5

Sandstone
Fig. 22- Cutters dsmaged

Fig. 19- Accaleratlons recorded at point G (181 rpm). by reverse rotation in the lab.

636

SPE 49204

TORSIONAL RESONANCE OF DRILL COLLARS WITH PDC BITS IN HARD ROCK

13

1300

1350

f 450

o~

10

is
Number of 30’ Drill

20

25

collars

,500

L——t——i l——+——l 20 0 100 200 0
RPM KLBS

l—+—+—v
400810$5

Fig. 23- Commonly observed collar vibration frequenclea at Catooaa.

Q’s

Fig. 25- Comparison of driling performance with steal and aluminum drillpipe.

8-1/2° PD~,
251
s q

18 DC,

Al

DP

60 rpm

20<0
~lo

120 rpm A 180 rpm

515.
10 20 30 40 50

Frequency, hz

~20 Drill collars

n*lo* 5/

.

/
.xM,,,,,..,..,:,::.,::.:

w =.,.:
---==:

~

/
<0

“~~&@+~g++: >
25 kips 30

. . . . . . .. . .. . .. . . . ... .. . . .

10

20 Frequency,

30

40

50

hz

10

15

20 Weight-on-bit,

Fig. 24- Comparison of tranafer function for aluminum drillpipe and steel drillpipe (0,1 damping, includes rig model).

Fig. 26- Affect of WOB and rotary speed on torsional resonance magnitude.

637


赞助商链接
相关文章:
更多相关标签: