ELECTRIC SUBMERSIBLE PUMPS
A method of pumping wellbore fluid, comprising the steps of: installing an electric submersible pump in a wellbore; and operating the pump at more than 4,500 rpm to pump the wellbore fluid. Pumping in this manner provides a number of advantages in use in that the required high-speed motor and pump is shorter for a given power than existing arrangements, and provides increased reliability due to reduced complexity. A much shorter motor/pump combination also allows such equipment to be used in deviated boreholes with a reduction in damage due to mishandling and bending, as well as facilitating assembly and testing in the manufacturer's plant.
This application is a divisional of U.S. patent application Ser. No. 10/562,255 (Atty. Dock. No. MRKS/0141), which is the National Stage of International App. No. PCT/GB04/02667, filed Jun. 21, 2004, which claims priority to GB 0314553.9, filed Jun. 21, 2003, which are herein incorporated by reference in their entireties.BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to motors and electronic drives for electric submersible pumps and compressors, and is concerned more, but not exclusively, with centrifugal pumps.
2. Description of the Related Art
Submersible pumping is a well-established technique for extracting hydrocarbons from deep boreholes, where the natural pressure in the reservoir is insufficient to lift the fluid or gas to surface. The technique is also used in water production.
Typically the production requirement is to lift large volumes of liquid against a pressure difference related to the depth of the well in which the pump is installed. For very heavy crude oils, slow-speed positive displacement pumps are suitable. These are usually rotated by a motor at the surface connected to the pump by a long flexible rod system. Centrifugal pumps have been found most suitable for normal crude oils, gas and water. These pumps are rotated by a submerged motor connected directly to the pump, with electric power being delivered from the surface by a long cable. Also, the use of electric cables makes installation possible in deep or long horizontal wells which would otherwise not be possible with the use of rotating rods.
The electric motors used for driving the centrifugal pumps are very elongated, sometimes of a length of more than one hundred times their diameter. The resulting complexity of such a device, the difficulty of its manufacture and the quantity of the degradable insulation materials it employs all reduce the system reliability.
Electric motor shaft power output is defined as the product of rotation speed and torque. For a given physical size and type of motor there is a limit to the level of torque that can be produced, typically due to self-heating. A high-speed motor therefore provides a means for obtaining more power from the same length of motor, or the same power from a shorter length.
The output of a pump is normally given in terms of its hydraulic power, which is the product of flow rate and lifting pressure (in rationalised units). Centrifugal pump technology is characterised by the power output being proportional to the cube of the rotational speed. This known relationship, sometimes termed the “affinity law”, means that a relatively small increase in the rotational speed can give rise to a substantial power increase.
Centrifugal pumps are frequently made with hundreds of impellers threaded on a common shaft, each impeller adding a little to the lifting pressure. Reducing the number of impellers by increasing the speed would therefore afford an improvement in reliability.
The above demonstrates that a high-speed motor and pump would, by being shorter for a given power, present direct advantages in reliability due to reduced complexity, or alternatively yield a higher output for a similar size. A large proportion of boreholes are deviated from the vertical and commonly even to the horizontal. A much shorter motor/pump combination would also lead to a reduction in damage caused by mishandling and bending during deployment through the curved sections of the borehole. Furthermore, the much-shortened length would allow motor/pump combinations to be assembled and tested in ideal conditions at the manufacturer's plant prior to being transported to the borehole location.
As will be described more fully below, innate limitations in the established motor and motor controller technology used in the electric submersible pumping industry have prevented the objective of higher speed being recognised or addressed.
Historically, electric submersible motors used for centrifugal pumping have been of the asynchronous, or induction, type. The stator is made of steel laminations and copper windings, and the rotor of steel laminations with copper bars forming the winding known as a squirrel cage. The rotor laminations are keyed to a shaft, this shaft providing the means of transmitting output torque. The rotor poles are produced by induction or transformer action between the stator and the rotor, using a portion of the stator current. The stator, in addition, produces a rotating stator field due to the alternating current in its windings. Since the transformer coupling to the rotor requires an alternating field in the rotor, the rotor must turn at a different (lower) speed than synchronous speed, producing a so-called slip frequency for induction. Electric submersible motors are made with two poles in order to achieve the maximum rotating speed from a standard 60 Hz utility supply. This speed is typically 3500 rpm, slightly less than the unattainable synchronous speed of 60 Hz×1 pole pair×60 s/min=3600 rpm.
It has become common to use variable speed drives to power these motors, rather than direct connection to the utility supply. Variable speed drives first convert utility AC power, typically at 60 Hz, to DC, and then by electronic switching convert the DC to a variable frequency alternating voltage. The use of a variable speed drive confers advantages during starting when it can limit the motor current to a safe level, and during production when it can be used to manage flow rates. The latter is important when the changing characteristics of a reservoir are considered over its producing life. Although variable speed drives, by creating an artificial supply of 70 Hz or more, can operate the motor at higher speed than when directly connected to the utility supply, this is a limited capability. Firstly the elongated induction motor is not suited to high-speed operations due to internal losses and small mechanical clearances, and secondly at the medium voltages used (often several thousand volts rms) drive losses become very high. Performance is generally limited up to 80 Hz or about 4500 rpm.
In order to maximise the induced rotor pole strength it is necessary to minimise the gap between the rotor and the stator. Unless very hot, the oil in the gap is sheared by the rotor turning yet remains in laminar flow. As a result the friction absorbs several percent of motor power. Motor efficiencies above 90% are sought, and this is an important source of loss in existing motors. The internal heating caused by these losses, and the copper losses in the squirrel cage, reduce motor life by aging the insulation materials.
The small gap is also a cause of premature failure due to mechanical causes. The limited diameter of boreholes is a natural disadvantage to both motors and pumps, and as a result their design is very elongated. A pump and induction motor assembly for producing 250 HP may be 20 metres long. This slender assembly is difficult to handle and particularly subject to damage when being deployed into deviated or horizontal wells, since small deflections of the motor housing can cause the rotor to impact on the stator. Rotor vibration due to bearing wear or imbalance also increases the chance of rotor impact.
The requirement for the rotor to be made of laminations and the limited overall motor diameter act together to constrain the diameter of the inner torque-carrying shaft. It is common practice, for example, to couple two 250 HP motors of 5.62 inch diameter together so as to make a longer 500 HP motor. Shaft strength limitation prevents this being increased to 750 HP or 1000 HP.
To provide a high-speed electric pumping system, it is desirable to increase the rotor clearance, and to reduce the internal sources of power loss that increase with speed. It is also necessary to use a drive technology which remains efficient at high speed and at the different operating voltage levels needed for different motor speeds required during the life of the well.
A further requirement of any high or low speed electric submersible system using variable speed drives is to minimise the deleterious effects of the electrical switching used to produce the alternating output voltages. Switching events on the long cables used in submersible cable propagate as wave fronts that reflect at connections and most particularly at the motor terminals. These reflections cause voltage transients that can approach twice the original voltage, and hence destroy insulation to earth. Commonly the motor voltage is presumed to be proportionally distributed through the turns of the stator winding, and the inter-turn insulation is less than that of the winding to earth. However a wave front impinging on a motor terminal must travel through the winding turn by turn before settling to its final value. Therefore there are short periods in which one turn of a winding carries the wave front at full voltage and an adjacent turn is unexcited. This internal voltage difference can exceed the inter-turn insulation rating, again causing premature failure. Increasing the insulation level to overcome these problems reduces the space available for the copper in the winding and also reduces the heat transfer from the copper, so that the motor specification is reduced.
An associated consideration for transients is the interference caused to data transmission systems used to convey data from instrumentation located in the well bore.
The foregoing has emphasised high-speed centrifugal pumping systems. However the same principles of reliable motor performance, matching efficient drives and circumventing the effects of transients on long cables are all applicable to positive displacement pumping systems.
Positive displacement pumps have a flow rate essentially determined by a characteristic volume per revolution multiplied by rotation speed. The torque demand at the pump shaft is determined by the back-pressure of the fluid column being lifted. These pumps usually operate at low speeds of a few hundred revolutions per minute. Since shaft power is the product of rotation speed and torque, it follows that these pumps are also characterised by extremely high torque demand. Where there is sand production with the pumped fluid and where the wells are deviated or horizontal, the rod connection to the surface has a very short working life. In these cases it is desirable to use a downhole motor with the positive displacement pump.
However, induction motors are inherently unsuited to low speed and high torque (although variable speed drives have improved their capabilities in this regard). Thus current installations rely on a gearbox to match the normal motor running speed and torque to the pump characteristics. This is also problematic as it is extremely difficult to make a reliable high torque gearbox in the small borehole diameter, and it is also expensive.
A motor having high torque at any speed including low speed is therefore preferable.SUMMARY OF THE INVENTION
It is an object of the invention to provide an efficient electric submersible pump, comprising a reliable electric submersible motor capable of operating at low, medium and high speeds, and overcoming many of the above-described disadvantages of existing motors.
It is a yet further objective of the present invention to provide a high power electrical submersible pumping system of the order of half the length of conventional equipment.
According to one aspect of the present invention, there is provided a method of pumping wellbore liquid, comprising the steps of: a) installing an electric submersible pump in a wellbore; and b) operating the pump at more than 4,500 rpm to pump the wellbore liquid.
It should be understood that references herein to “pumping of wellbore liquid” are intended to encompass within their scope the pumping of multiphase fluids, that is mixtures of water and/or oil and gas, as well as the pumping of wellbore liquid in a multi-lateral drilling environment in which the pump is operated to draw the wellbore liquid from a plurality of lateral well bores into a central sump.
According to further aspect of the present invention, there is provided an electric submersible pump comprising a permanent magnet motor having a rotor comprising a plurality of permanent magnets equiangularly spaced about a central shaft, a plurality of tubular elements supporting the permanent magnets spaced at different axial locations along the shaft, a retaining sleeve tightly fitted over the permanent magnets so as to retain the permanent magnets on the shaft, and a stator coaxial with the rotor comprising a stack of laminations and radially spaced coils wound around the stack.
The invention also provides a motor having a rotor comprising a carrier sleeve mounted on a central shaft, and a stator coaxial with the rotor comprising a stack of laminations and radially spaced coils wound around the stack, wherein the carrier sleeve is a loose fit on the shaft and is supported on the shaft by support rings tightly engaging the shaft.
The invention also provides a permanent magnet motor having a rotor comprising a carrier sleeve mounted on a central shaft and bearing a plurality of permanent magnets having axial ends, and a retention sleeve extending over the magnets and having at least one end turned in over at least one stress-relieving radially outwardly extending abutment part on the carrier sleeve abutting an adjacent axial end of the magnets to retain the magnets in position on the carrier sleeve without damaging the axial end of the magnet.
The invention also provides a permanent magnet motor having an elongate rotor provided with elongate permanent magnet means extending therealong, and a stator coaxial with the rotor, wherein the permanent magnet means incorporates axially laminated parts to reduce eddy current losses.
The invention also provides a motor having a rotor and a stator coaxial with the rotor, wherein the rotor is mounted in a bearing, and one of the stator and the bearing is provided with resiliently biased projection means for engaging within receiving means provided on the other of the stator and the bearing to prevent relative rotation therebetween when the rotor begins to rotate with respect to the stator on starting of the motor.
The invention also provides a motor having a rotor and a stator coaxial with the rotor, wherein the stator is mounted in a housing, the stator being locked within the housing by an axial key engaging within an axial groove in at least one of the stator and the housing to prevent the stator from turning relative to the housing.
For a better understanding of the present. invention and in order to show how the same may be carried into effect, reference will now be made, by way of example, to the accompanying drawings, in which:
With reference to
A power cable 110 for the motor 108 is run up past the pump 107 and alongside the tubing 105 until it emerges at the surface wellhead and passes to a variable speed drive 111. This drive 111 is powered by the utility supply 112 or a generator.
It will be appreciated that other configurations of the installation are possible, such as mounting the pump below the motor, and taking the cable up the tubing or making it an integral part of the tubing. Arrangements, such as that disclosed in U.S. Pat. No. 6,000,915, which accommodate the pump concentrically within the motor bore will generally be found to make poor use of the limited borehole cross section and are not preferred.
During high-speed rotation the magnets 204 experience considerable centrifugal force, and the adhesive that bonds them to the sleeves 203 may weaken with age. A retaining sleeve 205, preferably of metal, provides a durable means of retention. To avoid the use of materials which degrade during prolonged operation at high temperature, it is preferable to make the sleeve 205 a tight fit by shrinking it on, rather than depending upon tape, adhesives and fillers. The sleeve 205 is preferably of one piece although, for ease of assembly, several shorter rings may be fitted adjacent to each other if required. U.S. Pat. No. 4,742,259 discloses a technique for fitting a sleeve with axial constraint. This technique requires the fitting of end washers that are pressed to the shaft to locate them without using positive abutments to do so. In a preferred arrangement shown I\N
The assembly so far described is termed the rotor, and the length of motor delineated by a sleeve is termed a rotor stage.
The magnets 204 are circumferentially disposed about the sleeve 203, and alternately poled in an essentially radial direction to cause aspatially alternating magnetic flux to cross the clearance gap 209. Other magnet arrangements will be known to persons skilled in the art. The entire motor and hence the gap 209 are filled with a benign fluid, such as a highly refined mineral oil, to balance the inside of the motor against the external wellbore pressure.
Preferably the magnets 204 are plated, for example in a vapour deposition process, with corrosion-resistant material such as aluminium, so that they may resist corrosion from any ingress of moisture into the motor or from other sources, and so that any small loose particles of magnet material will be sealed into the magnets and not come free to circulate within the motor bearing system.
Contained within the tubular housing 202 is a stack of thin magnetically permeable laminations 206 as may be seen more clearly in
The PMSM motor constructed as described will have many desirable characteristics for submersible pumping, associated with the general nature of permanent magnet motors. For example, by providing a rotor flux from permanent magnets, there is no need to energise the rotor, unlike the field winding that requires separate power in an induction motor. This reduces the motor current by the amount needed for rotor magnetisation, which therefore reduces the ohmic loss in the stator windings and the power cable. It also eliminates the rotor cage winding and thus an internal source of heating. The copper within the stator is used only for the production of the rotating stator flux. The inherent torque output for the motor, which is derived from a product of space utilisation, rotor flux and stator flux, is very high compared to an induction motor. This torque is available at any speed.
Further aspects of the motor construction may be addressed to give reliable high-speed performance. Firstly, as mentioned above, a major source of inefficiency in induction motors is the frictional drag in the necessarily small rotor-stator gap. In a normal mass-produced PMSM the gap is also kept small in order to economise on the amount of magnet material required. However, if it is considered that permanently magnetic material is not itself significantly magnetically permeable, then for magnetic purposes the gap between the stator and the rotor is that between the lamination tip 210 and the outer surface of the sleeve 203. The mechanical clearance gap 209 is only a part of this. Thus, if, for example, the magnet thickness was 3 mm and the clearance gap was increased from 0.25 mm to 1.25 mm, or 500%, the magnetic gap would only have increased 30%. With only a modest increase in the amount of magnetic material it is possible to purposely design the motor to ensure a sufficiently large mechanical clearance such that at high speed the fluid in the clearance is turbulent. Above 5400 rpm for a rotor of diameter more than 50 mm a gap greater than 1.25 mm is preferred for this purpose. A designer may use the known Reynolds number theory to estimate the needed gap size for other operating conditions, fluids and motor sizes. Although the friction loss is higher in turbulent flow than in laminar flow, turbulent flow ensures much more effective heat transfer between the rotor and the stator, so reducing the maximum internal temperature. At any speed the large clearance will reduce the likelihood of mechanical damage to the rotor during installation caused by bending of the outer housing, and also provide a measure of tolerance to contaminant particles.
Furthermore, it will be found that the deliberately large gap reduces the eddy current losses, and hence heating, induced in the retaining sleeve 205 and magnetic material according to their conductivity. These losses increase approximately as the square of rotation speed, but diminish with distance from the lamination tips 210. The inter-magnet spaces 213 may be filled but, unless care is taken to seal the cavity, particles of filler may dislodge over time and damage the motor. If required the cavities may be left unfilled. This is made possible by the sleeve 205, since it presents a low drag rotating surface to the clearance gap while making an enclosure to trap the fluid in the cavities. This trapped fluid is limited to bodily rotation or axial flow and does not contribute to friction in the clearance gap.
A further reduction in eddy current losses in the rotor can be obtained by laminating the magnets axially. Rotor eddy current losses originate from flux harmonics in the stator, the eddy currents circulating on the face of the magnets and penetrating through the depth of the magnets and then into the steel that the magnets are bonded to. Most of the ohmic losses resulting from this current flow are in the magnets, assuming the retaining sleeve is non-conducting or very thin, and the current flow increases with the face area of the magnets. Accordingly, in the same way that the stator steel is laminated to reduce the effect of rotor flux, the magnets can be laminated to reduce the effect of stator flux. For an elongated motor the face area is the width of the magnet times the continuous length of the magnet section. Therefore, by using a series of short magnets to make a continuous length that are electrically isolated from one another where they would otherwise touch, the effect is to produce an axially laminated magnet. Practically the magnet ends may be coated with epoxy or varnish during assembly or spacers used. An approximately equal length and width of each magnet will be found to give a good reduction in losses while not unduly complicating manufacture. This method is unlikely to work usefully in motors of conventional length to diameter ratios as the magnet face area is already relatively small.
Other types of PMSM construction are possible, while maintaining the large gap. A slot-less construction in which the laminations become a stack of rings, or are replaced with a magnetically permeable tube, requires much more magnet material and will normally be found uneconomic for submersible pumping.
It is also possible to design the PMSM so that the slots are fully closed or almost fully closed in the vicinity of the lamination tips 210. This ensures the retention of the winding without use of insulating retaining wedges that may degrade. It also reduces the cogging torque, that is the alternating accelerating and retarding torque developed as the magnets come into and out of maximum overlap with the teeth.
For the purpose of maximum power output at high efficiency it is necessary to optimise the electromagnetic design. Unlike conventional submersible pump induction motors, which invariably have two poles for the reasons given above, it will be found that the optimum number of poles is usually six or more for PMSM motors up to approximately seven inches (17.5 cm) in diameter. Four poles will give an acceptable output for smaller motors but even more poles are preferred for larger motors. The higher pole count allows the flux density in the stator laminations to be better distributed so that the amount of steel in the outer areas 211, may be reduced. This permits the area of the slots 207, and hence the amount of copper in the windings, to be increased. When the high frequency restrictions discussed below on drive output are considered it will be appreciated that, in larger motor sizes, higher pole counts are more demanding of the drive. A limit may be reached where accepting additional stator lamination outer material is appropriate to make the drive practical. Conversely as taught in U.S. Pat. No. 6,388,353, with drives and step-up transformers typical of oilfield induction motor technology, a high pole count motor permits operation at low speed and high torque for progressive cavity pumps. For example, a ten-pole motor driven at a frequency of 60 Hz will rotate at 720 rpm.
U.S. Pat. No. 4,513,215 and U.S. Pat. No. 4,521,708 teach means added to the bearing outer ring for pegging or gripping the bearing outer ring to prevent rotation during motor start up, before thermal expansion has taken effect. However the larger shaft diameter made possible with PMSM motors necessarily reduces the bearing outer ring wall thickness so that such known methods cannot be used with such motors.
An improved method of assembling the rotor illustrated in
The need to assemble the rotor with long solid sleeves 203 presents a problem in that a stable fit to the shaft 201 is necessary, but the required shaft straightness for closely fitting sleeves 203 to pass smoothly over the shaft 201 during assembly is very demanding. The preferred means of assembly is to use support rings 411 as shown in
The use of a relatively short key, such as the key 413, ensures that the torsional stress in the sleeve 203 is limited to that caused by the torque generated by the magnets on the same sleeve 203. In a long motor, the portion of the shaft 201 under the sleeve 203 nearest the output end of the motor will carry a high torque accumulated from all the other sleeves. Particularly where multiple motors are connected in series to increase power output, this torque can be very high. If the accumulated stress in the shaft 201 were to be shared with the sleeve 203 by way of a long key, there would be a risk that the magnets, being brittle, would fracture. (In submersible induction motors long keys are used to maintain all the laminations of the rotor in alignment, as well as to transfer torque.)
A further consequence of very high torque is that twist in the shaft 201 may cause sleeves 203 at opposite ends of the shaft 201 to come out of alignment with each other and hence with the stator, with the result that the sleeves 203 cannot at the same time produce maximum torque. U.S. Pat. No. 6,388,353 suggests mounting the sleeves on the shaft with an angular skew relative to one another so that, when twisted in use, the sleeves are brought back into alignment. Alternative methods that can be used with series-connected motors are (i) to stagger the angle of each shaft to the next in line, such as by cutting the splines at the ends of the shaft with a small angular offset relative to one another, or (ii) to connect the housings to one another with a small angular offset. Within a single stator, the simple expedient of twisting the stator will effect a compensating correction to a twisted shaft, and is similar to a well-known technique for reducing motor cogging torque. The compensation of all these methods has variable effectiveness as the skew is fixed at one angle to compensate the angle of twist at one level of torque, and cannot therefore be correct at other levels of torque. It should be noted that the amounts of twist referred to are very small, typically less than a degree, and the problem may not be significant if the motor is designed to maximise shaft diameter and hence resistance to torsion. Accordingly it is preferred to design elongated motors so as not to suffer from excessive shaft twist.
The sleeve 403 carries, or is integral with, the shaft bearing. A ring 407 of bearing quality material, such as that marketed under the tradename Deva Metal, may be pressed onto the ring 411. The outer ring 420 of the bearing runs on the ring 407, and is axially captured by thrust washers 421 which themselves are captured between the sleeves 203 and the support rings 411. Alternative arrangements for the bearings, in which for example the support ring 411 is made entirely of bearing material, eliminating the ring 412, are possible within the scope of the invention. Similarly the outermost rings 411 may be of modified shape as their outermost ends do not mate to rotor sleeves.
Substantial heat, of the order of 100 W, will be generated within the bearing. This heat is transferred to the laminations and thence the motor housing by two main means, namely conduction through the outer ring 420, and conduction through the support ring 411 and the rotor. In the latter case heat passes through the magnets near each bearing and across the oil-filled rotor/stator gap. This second path, though less direct than the first, will significantly raise the magnet temperature. A thermal insulator in this path between the bearing running surface and the magnet, such as may be provided by making the support ring 411 of ceramic, will increase the thermal resistance in this path, and thus reduce the magnet temperature rise.
In a completed PMSM the rotor is centred by the bearings 401 and so is magnetically balanced. Particularly when the motor is installed vertically the bearing loads will be very low, and the bearings 401, which necessarily run hydrodynamically for maximising lifespan, may become unstable, resulting in shaft whirl and other vibrations. In the present invention therefore, the bearings 401 must be designed to create sufficient internal pressure to remain stable and hydrodynamic at low shaft load.
A further means of purging oil through the bearings 401 is to bore the motor shaft 201 for the introduction of oil throughout the length of the motor to cross bores 406 at each bearing 401 as shown in
If the stator bore is of constant, carefully controlled diameter, then the rotor assembly, complete with bearings, may be slid into the stator. The stator thereby provides the outer support for the bearing outer rings 420. However this known arrangement necessarily requires the bearing rings to rub the stator bore during insertion, with the possible risk of abrasive damage.
Alternatively the stator bore may be made of smaller diameter 403 in the axial sections opposite the bearings, such that the reduced bore lies within the stator to rotor clearance, as indicated by the broken lines 408 in
This means of assembly is not immediately suited to PMSMs due to the extremely high side magnet forces between the rotor and the stator caused when the rotor becomes slightly eccentred in the stator bore. This is well known in small industrial motors where external fixtures are used to handle the forces involved during insertion. In an elongated motor the problem is very serious since it is not effective to support the shaft from each end. It may be found from electromagnetic calculations that a force of thousands of Newtons per metre of rotor length and millimetre of deflection may be produced. A much smaller force is sufficient to bend, or deflect, a shaft held only at its ends. Any deflection increases the side force leading to more deflection. Thus if the rotor bearings mate only to a restricted stator bore, then for most of the insertion process they will not provide support to the rotor, and the rotor will deflect until the bearings touch the stator bore between the restricted sections. It would be difficult to slide the rotor into the motor, and also the bearing outer diameters will now be offset from the restricted bore section 403, so that the restrictions become obstructions.
The preferred embodiments of the present invention radically change the method of assembly of elongated motors. Firstly the more conventional approach to making small non-submersible motors is adopted, in which the stator is wound before insertion into its housing.
A shrink fit of the finished stator is used to ensure high contact pressure with the housing, reducing thermal contact resistance and thereby minimising the internal temperature of the motor. However the invention provides for the special circumstances of an elongated motor that has internal bearing surfaces that may be on a reduced stator bore diameter.
The laminations may be welded together on their outer diameter so as to maintain the close stacking of the laminations, or some other known means may be used. After winding and possibly varnish impregnating, the stator may be ground on the outside diameter to make it a close fit in the housing. Preferably the. housing will be pre-expanded in order that, after insertion of the stator, the housing will relax to a shrink fit. The pressure of contact will then greatly reduce the contact thermal resistance. A means of expanding the housing for assembly is to pre-heat it. Conversely, to repair a stator, or to recover the housing for re-use, it will be necessary to expand the housing with the stator in situ. The length of the stator renders the required force to be too high for a press tool to be used. A preferred means is to use an induction heater, which essentially comprises an electrical coil that slides over the housing, the coil being connected to a power generator of appropriate frequency. It is known from the theory of induction heating that, by choosing a frequency such that the skin depth of radiation in the housing does not penetrate through the housing, energy may be rapidly and selectively imparted to it without penetrating the stator. This provides a time window in which the housing will release the stator and the stator may be extracted before it becomes heated by diffusion and expands to re-establish the lock. This method is suited even to elongated motors.
A further method with great advantages in terms of manufacturing and repair cost, as well as in terms of reliably preventing the stator from rotating within its housing, is to mechanically lock the stator to the housing with an anti-rotation device or devices. One possibility is to provide an axial key 3603 between the stator and the housing 202, as illustrated in
The embodiment disclosed has the further advantage over conventional construction of induction submersible motors that, by providing easy access to the end-windings, the highest quality winding procedure may be followed and the results easily inspected. This technique is applicable to all types of elongated motor.
One advantage for reliability is that there are only three coils, one per phase. Consequently there is no crossing of phase windings at the end turns. The end turns 2205 possess further advantages. The end turns 2205 fall naturally within—the radial limits of the slots and are short. This minimises risks with insulation chafing, provides a short path back to the stator for conduction cooling, and has minimal energy waste in unproductive copper. This in direct contrast to double-layer lap windings, as used in induction submersible motors. In these the end turns need to expand beyond the radial limit of the slots and/or become very long in order to accommodate the wire crossings between layers and phases. Since the winding area is constrained by the housing internal diameter and the bearing/rotor outer diameter, the problem is severe.
A winding with six slots and eight poles as described above provides phase separation in the slots. However, at high speed, the high pole count makes heavy demands of the variable speed drive as described below. Furthermore the self-inductance of each winding is high, requiring more drive voltage to overcome its reactance for a given motor current, which is manifest as a poor power factor. A compromise embodiment of six poles and nine slots for a motor having three phases A, B and C, as shown in
A preferred improvement to the laminations where two coils are adjacent in the same slot is to introduce partial teeth 3110 shown representatively in
A further, preferred feature, that may be used where the coils are pre-fabricated and accommodated in a common slot including partial teeth 3110 for separating adjacent coils as shown in
A further, and preferred, aspect of the present invention is shown in
The outer lamination stacks may instead be made as magnetically permeable tubes of cast insulated iron powder, with the advantages of offering a smooth surface where the inner and outer parts of the stator come together, and of economy of materials. Each tube may be made by combining smaller arcuate segments, to reduce the size and cost of the casting, such an arcuate construction being unfeasible with laminations which would effectively be small steel fragments. The partial teeth disclosed above may be incorporated into the outer ring of the split lamination. The disclosed multi-part stator must also be multipart or open-slotted outwards at the bearing sections 202 to permit loading of the winding from the outside.
This method of assembly translates the known advantages of form-wound coils used in physically large industrial motors to the difficult elongated small diameter geometry of submersible motors.
Form-wound coils for lap wound large motors are manufactured separately from the laminations and are then inserted into rectangular slots open to the stator bore. In submersible motors there is insufficient working space in the bore of the laminations to load the formed-coils ready to insert into the slots, and for high speed motors the open slots would cause substantial losses.
By opening the lamination slots from the outside, the invention permits formed coils to be used while not incurring these problems. —The particular advantage of the concentrated winding is that the formed coils are very simple and that the large winding slots for the small size of motor facilitates the use of semi-rigid rectangular or wedge-shaped copper wire.
With formed coils the wire is bent once at any position as it is wrapped to form the coil, unlike the conventional process for winding elongated motors in which the wire is threaded back and forth through the slots. Consequently a much more rigid wire may be used, as there is no work hardening and insulation damage that would occur if it was attempted to wind conventionally, with repeated bending. Rigidity solves the known problem in elongated motors of wires crossing within a coil deep inside the stator. It provides a non-rubbing and stiff end turn assembly. Round wire is known to give a very poor copper fill factor in a slot compared to rectangular wire, essentially because the latter packs together better. Typically the thermal conductivity from the copper through its insulation back to the lamination is improved also. With a high copper fill the motor will have much reduced internal heating compared to a conventionally round wire wound motor. This is a source of improved reliability, or alternatively of higher torque for the same temperature rise.
While rectangular wire is preferred, it will be appreciated that formed coils made from round wire will nevertheless be superior to round wire conventionally wound by threading through a stator. For a lower pole count motor, such as the commonly used two-pole induction motor, the winding pitch is necessarily substantially half the circumference of the motor. This means that many end turn crossings are unavoidable.
The flexibility of round wires is beneficial in this case, whilst retaining the key advantage disclosed above of prefabricating the coils.
Formed coils that are fully encased in insulation over the portions that enter the stator slots do not require insulating slot liners. Furthermore it is not necessary to impregnate the coils after insertion to complete the insulation and secure them into the slots, provided that the insulation is impregnated or encapsulated prior to insertion and the slots are shaped to retain the coils as disclosed above. The means that the present invention not only allows the stator to be removed from the motor housing, but permits the winding to be disassembled from the stator. Axial movement of the coils can be prevented by inserting insulating blocks between the end turn loops and the stator.
Despite the many advantages of the split lamination construction, it does require careful design and attention to manufacturing to ensure satisfactory engagement of the stator parts. All the aforementioned advantages for windings may be obtained when the number of coil turns is not too large, and especially for short-pitched coils, by using one-part laminations as will now be described with reference to
Referring again to
The stator sections may be carried and aligned on a common mandrel for insertion in the motor housing 202, similarly to the foregoing descriptions for a single stator. The stator bore restrictions in which to house the bearings 401 are replaced by housings 404 concentric with the individual stator sections. Concentricity is maintained by the motor housing 202 when the entire assembly of bearing housings 404 and stator sections are inserted. The series connection 405 of the windings 214 of the stator sections is preferably achieved by permanent means such as brazing. The use of connectors, while possible, reduces reliability. It is a feature of the invention that winding before insertion permits these connections to be made and inspected beforehand.
High speed multi-pole PMSMs present. a variable speed drive problem that the present invention addresses as described below. The origin of the problem is that the base electrical frequency that the drive must generate is the product of the number of motor pole pairs and the number of shaft revolutions per second. A standard induction motor having two poles and turning at 3600 rpm therefore has an electrical frequency of 60 Hz. A PMSM in accordance with the invention rotating above 4500 rpm has a much higher frequency. At 7200 rpm and six poles, the electrical frequency is 360 Hz. This six-fold increase is a step change in operating conditions for electric submersible pumping systems and well beyond the range of general industrial drives.
It will be appreciated that multiple motors, or stators, may be connected electrically in series so that the resistances, inductances and EMFs add to make a single equivalent larger motor with a common shaft. Placing the terminals in parallel is also possible but poses difficulties in controlling currents between all the windings. More realistic motor models in which for example the EMF source and inductance are lumped together as an element that calculates the time rate of change of the internal flux linkage, and in which magnetic saturation is taken into account, are all refinements which do not affect the present invention. The number of phases, three, is well suited to the task of electric submersible pumping motors, but is not limiting.
An idealised PMSM as described with reference to
The sinusoidal nature of the electrical quantities is ideally suited to the task of electric submersible pumping. This is because the smoothly varying waveforms do not cause damaging transients at the motor terminals, and because the motor torque can be shown to be constant with rotation, which reduces the likelihood of torsional vibration.
The construction of such a motor requires careful attention to the distribution of the turns of the windings within the stator slots. To produce a sinusoidal EMF with a reasonable number of slots cut in the laminations requires the turns from different phases to share slots and to be distributed among many slots. This immediately causes a reduction in reliability due to the potential for insulation failure in the many end-turn crossings and due to the mixed phases. There is also the loss of useful copper due to increased insulation between the phases.
When the windings are made so that the phases are kept in separate slots, the back EMF will be more similar in form to E in
By contrast, in a brushless DC motor inverter, the filter is not present and the motor is connected directly to the terminals AA, BB and CC. Only two phases are active, that is only one switch is turned on, at a time, whilst the switches for the third phase both remain turned off as described above.
In driving of the motor in accordance with the present invention all three phase outputs are active at all times. In a sinusoidal variable speed drive it is necessary to use pulse-width modulation (PWM) or other switching modulation scheme known in the art, e.g. hysteretic, space vector, switching table, to create the effect of a sinusoidal output current. In the following description. PWM drive is referred to by way of non-limiting illustration.
However, to produce a high-power high-speed variable speed drive with sinusoidal output presents severe difficulties, as will now be described. The method is best suited to trapezoidal or similar EMF but is also applicable to sinusoidal driving, the difference being in the harmonic content of the waveforms and hence the best use of available power capacity.
The majority of variable speed drive circuits operate at typical utility supply voltages of 380 V AC-690 V AC, since the power semiconductors that they use for switches are well proven and efficient. However, just as in utility power transmission, for efficient motor operation using long power cables, it is necessary to use Medium Voltages, commonly in the range 1000 V AC-4000 V AC. Such voltages reduce the motor current and hence the ohmic losses in the cable. The majority of variable speed drive circuits for use with submersible pumps are therefore installed with a step-up transformer on the output. These transformers are a source of additional power loss, direct cost, and are often large and oil-filled, requiring special environmental precautions and substantial space. A wide speed range requires expensive core material for high speed but also a very large core to prevent magnetic saturation if operation at low speed is also required. They are in addition to input transformers, commonly required as described below to reduce harmonic distortion of the supply and to match to the available supply voltage.
A Medium Voltage drive circuit operates from a supply voltage directly at the voltage which is required for the motors. It therefore eliminates the undesirable output step-up transformer but has certain limitations for the purpose of high speed pumping.
Medium Voltage power semiconductors when used for the switches 902 and 903 of the drive circuit of
To produce the quality sinusoidal waveform in
With high speed multi-pole motors the switching speed becomes too high to be economic with Medium Voltage semiconductors. For example, a high speed motor with six poles operating at 7200 rpm has a fundamental frequency of 360 Hz, so that the drive should operate with a switching frequency of at least 3600 Hz just to achieve the quality of the response of
The present invention overcomes these problems by deliberately over-modulating (non-linearly modulating) the output stage of
The distorted switching waveform shown in
Since the amplitude of the drive voltage is fixed once over-modulation is employed, the only way of varying the motor voltage and hence the speed is to vary the internal drive voltage Vjick applied between the terminals 904 and 905 in
However, the present invention seeks to make use of the special characteristics of the pumps it is powering, in order to further improve drive performance. In accordance with the power characteristic of centrifugal pumps as mentioned earlier and as depicted. in
Therefore a properly specified drive can be assumed to be run most of the time above half speed.
A first embodiment of adjustable voltage converter 113 particular to the present invention incorporates a specially adapted variable voltage chopper source shown in
In this circuit a first fixed supply voltage source 1401 is connected in series with a second fixed supply voltage source 1402, and a chopper, comprising a switch 1403, a diode 1404, an inductor 1405 and a capacitor 1406, is connected across the source 1402. By varying the duty cycle of the switch 1403, the voltage across the capacitor 1406 may be varied between zero and the fixed voltage of the source 1402. Since the voltage across the capacitor 1406 is in series with the fixed voltage of the source 1401, the voltage across the output terminals 904 and 905 may be varied from the fixed voltage of the source 1401 to the sum of the voltages of the sources 1401 and 1402.
When the motor is operating at low speed, as when starting and stopping, the power level and motor frequency will be low. Consequently conventional pulse width modulation by the drive output may be used with little penalty, with the chopper turned off, leaving the drive voltage fixed at the level of 1401. At full power output the chopper may be left permanently on. Therefore it has no switching losses in either case.
The switching losses of a chopper are proportional to switching frequency, input voltage and output current. The advantages of the arrangement shown are that the input voltage is only half that of the full supply, and that the frequency of chopping may be set independently of the motor speed since it is used to produce the link voltage and not the modulated drive output to the motor. For example, with a pump load and high speed corresponding to high power, the chopper might be operated at 500 Hz to limit switch losses, whereas the output stage in
If a chopper were used across the full available supply voltage, the losses would be doubled as the switching voltage would be doubled while the switched current remained the same. This may be acceptable for lower power low cost drives where the dual fixed voltage supply 1401,1402 is not implemented.
A further feature to optimise the drive, based on the characteristics of the electric submersible pump. is to vary the chopper frequency. Inductor 1405 is heavy, costly and has losses proportional to ripple current and average output current. As such it is an undesirable addition by comparison with conventional drives.
The inductance value is usually chosen to limit the chopper ripple current to a reasonable level. The ripple current is at a maximum when the chopper output is half its input voltage (50% duty cycle). At the same time, because of the nature of the pump load, the output power will be significantly reduced. Therefore at this condition the chopper frequency can be relatively high, permitting a lower inductance value. As the voltage increases the output power will increase, and the chopper frequency must be reduced to limit the switching losses. Since the ripple reduces as the output voltage increases (higher duty cycle) but increases as the frequency is reduced, it can be. seen that a compromise profile of frequency versus power output can be found which allows a much smaller value of inductance than would otherwise be the case, reducing the adverse factors of weight, cost and power loss. It is quite reasonable to reduce the value by a factor of two, or four if the chopper is connected across the full supply and not a portion of it. Thus variation of the internal frequency of the adjustable voltage converter with output serves to improve efficiency and/or reduce the size of components.
A particularly beneficial choice is when one secondary winding is wye-connected and the other secondary winding is delta-connected, and the turns ratios are adjusted such that the rectified outputs are equal. In this case the current pulses taken from the supply from one capacitor are displaced in time with respect to the current pulses taken from the other capacitor. This known arrangement is beneficial to the supply as the current pulses taken from the supply by the assembly occur twelve times per supply cycle and not six as when using a single rectifier. This substantially reduces the harmonic distortion imposed by the drive circuit on the utility supply.
A further embodiment of adjustable voltage converter 113 in accordance with the present invention incorporates a three-phase boost converter as shown in
Other arrangements for efficient variable voltage are possible. If the fixed voltage sources 1401 and 1402 are kept separate and a chopper is placed across each, and the chopper outputs are connected in series, one chopper may be kept fully on or fully off which the other chopper varies its output. In this way the entire voltage range may be covered with only one chopper operating at less than the sum of the available supply voltage. It is also possible to connect the chopper across the fixed voltage source 1401 and to connect the fixed voltage source 1402 separately to the chopper output so as to add to it. Where the interference to the utility could be tolerated, one or both of the rectifiers in
It is desirable to dynamically control the drive, cable and motor arrangements to realise optimum efficiency. Unlike an induction motor a PMSM must be driven synchronously with its shaft rotation. With a brushless DC motor, for which two phases are driven at a time as hereinbefore described, there are numerous schemes to determine the shaft rotation without the use of rotation sensors (such as “Sensorless Vector and Direct Torque Control”, 1998, P Vas, Oxford University Press) based on observing the effect of the voltage on the undriven phase. With a PMSM, because all three phases are continuously driven, the shaft rotation must be determined in another way. In a factory, a means is required to directly measure the instantaneous angular position of the rotor using a shaft sensor, such as a resolver or encoder, and to use the result to control the phase of the voltage or current output of the variable speed drive circuit. However, apart from the uncertain reliability of such transducers, the additional cabling or other means needed to transmit the position information from deep in the borehole to the variable speed drive circuit at the surface makes a sensorless technique almost certain to be required.
It is possible to control AC PMSM motors without sensors utilising a computerised or discrete component model of the motor, based on an electrical equivalent circuit as in
The characteristic of the load that is required for the feature of the invention now to be disclosed is that its power be steady for a steady speed. By averaging over a sufficiently long period random or short term load fluctuations can be accommodated. A submersible centrifugal pump, or a turbine, meets this condition.
Therefore, if. the PMSM motor control is changed while keeping the frequency, and hence synchronous speed, constant, the load power will remain unchanged. The optimum control condition will be when the drive output power, which is measurable, is minimised. For example if a pump is turning at a fixed speed and takes 300 kW from the drive at one control condition, and 298 kW at another, then the second condition is more efficient as there is less power supplied, regardless of what the motor and cable parameters are thought to be and regardless of what the actual pump power is, since it has not changed.
Suitable means of effecting these control changes to find the most efficient system operating point are now disclosed. At constant speed the internal mechanical friction losses will be fixed, so that the dominant variable loss that needs to be reduced to a minimum by the control is the ohmic loss in the windings.
The EMF E depends to a good approximation only on the motor speed, so that, for fixed speed operation and since the output power is fixed by the load, the quantity I cos (φ) will be constant. The broken line 1806 shows the locus of constant output power. It is evident that minimum current, and hence least loss in the resistance of the copper winding, ≈I2R, occurs when φ is zero.
In open-loop operation of the PMSM motor a given three-phase voltage or current is applied at a given frequency. The motor operates in accordance with the phasor diagram at an angle, satisfying the relationships between the parameters. Operation is at some non-zero 0, and as a result the motor is never optimally efficient. If+ increases beyond approximately 45 degrees, the motor operation becomes unreliable since fluctuations in the load may increase the angle to the point where there is insufficient current available from the drive circuit to maintain the output power. At this point the motor will lose synchronism and stall. If the conditions are as shown in
Therefore the optimum operating point of the system at a given speed may be found by varying the input voltage independently of particular knowledge of the motor or cable parameters or actual load power. Since the motor output power demanded by the pump for constant pump speed and fluid type is constant, the control observable is the input power measurable at the surface. Parameters corresponding to the age or nature of the motor and/or cable are not required.
This method has broader applicability. For example, if a current-controlled drive is used, then current may be used as the control variable and the drive voltage and output power will vary as a consequence. Alternatively, for a fixed amplitude of voltage or current, the phase of this quantity relative to the estimated rotor angle could be slightly changed. In this case the optimum condition will be when the speed is maximised, since, from the characteristics of the load, increased speed corresponds to increased output power, and the maximum output power always occurs when reaches zero. (In
Practically the above method is applicable to PMSMs without the simplified assumptions of
The improved winding and construction methods described above make it possible to further extend the reliability of the entire pumping system by means of cooperative duplication of failure prone electrical elements. The cost of replacing a failed system, and the loss of fluid production until the repair is complete, will far exceed the cost of the duplicate parts to be described.
The advantage of such an arrangement is that failure of the cable 110 or its splices or connections, or of the drive circuit 111 or of any of the coils of the winding set 3001 will not affect any of the corresponding parts associated with the winding set 3001′. By using concentrated windings the windings of the phase sets may more easily be kept well insulated from one another than with lap windings. In the simplest case one coil per phase is wound over alternate teeth in twelve slots, the corresponding phases of each set being adjacent to one another. In this way the six motor leads exit the motor directly from their coils without crossing.
The above arrangement of six phases split into two sets of three phases, though having practical advantages, is not limiting. However two-phase motors still require three conductors in a cable, whereas motors with a larger number of phases require further cable conductors, which is undesirable.
A six-phase, drive output circuit may be constructed by adding three extra pairs of switches to
In an alternative embodiment of the fault-tolerant pumping system, two motors are connected mechanically in series on a common shaft, but powered by separate cables and drives as before. The two motors may be operated individually or simultaneously. In the latter case the drive circuits must again be arranged to cooperate. This method is not applicable in the majority of wells as usually the motor diameter is the largest that can be fitted within the casing 103, and the motor cable 110 is fed past the pump and into the top of the motor 108 immediately below it. In such cases it is not possible to pass a second cable past this first motor to a further motor arranged at a deeper level.
The foregoing has assumed two duplicate motor sections that are identical, as that is the simplest fault tolerant arrangement. However, with suitable changes to the control and drive levels, a plurality of motor sections cables and drives of different characteristics may be used within the scope of the invention, so long as they are controlled to the same shaft speed.
The electric submersible pump system of the present invention has broad application, particularly in the field of downhole wellbore operations. Drilling for wellbore fluid at large depths is typically restricted to relatively narrow boreholes, so the facility of the present invention to provide the same motor power in a smaller overall package is immediately advantageous.
A further application of the present invention is to compress wellbore fluid in situ. It may sometimes not be required to immediately transport the wellbore fluid to the surface from its underground reservoir, but to compress it either for later recovery or merely to facilitate further exploration. Alternatively it may be required to transport the wellbore fluid from a first subterranean location to a second subterranean location, for the above reasons amongst others.
A recent development in mining operations is the application of multi-lateral wellbore systems in which a number of small diameter wellbores are drilled substantially horizontally from a central subterranean sump. Currently known pumping systems have significant difficulties in pumping from lateral wellbores, whereas the pump of the present invention can still maintain a high output in such environments. In this case, wellbore fluid is transported from the multiple lateral wellbores to the central sump, where it may be recovered to the surface or compressed as described above.
As hereinbefore described, an objective of the present invention is to provide a high-speed electric submersible pump, capable of operating at speeds above the current maximum of approximately 4,000 rpm. The standard operating speed of embodiments of the invention intended for the above applications is above 4,500 rpm, and an optimal speed, providing a marked improvement over current systems, is approximately 7,200 rpm.
The present invention discloses a permanent magnet synchronous motor submersible pumping system. It will be appreciated by the person skilled in the art that various modifications may be made to the above embodiments without departing from the scope of the invention.
Reference should also be made to “The Technology of Artificial Lift Methods”, Vol. 2b, K. E. Brown, Penwell Publishing 1980, the contents of which are incorporated herein by reference.
1. A permanent magnet motor comprising:
- a rotor, comprising: a central shaft; and a plurality of permanent magnets having axial ends; a carrier sleeve mounted on the shaft and bearing the magnets; a retention sleeve extending over the magnets; and at least one stress-relieving radially outwardly extending abutment part on the carrier sleeve abutting an adjacent axial end of the magnets, wherein the retention sleeve has at least one end portion turned in over the abutment part to retain the magnets in position on the carrier sleeve without damaging the axial end of the magnet.
2. The motor according to claim 1, wherein:
- the rotor comprises a second outwardly extending abutment part on the carrier sleeve abutting an adjacent axial end of the magnets, and
- a second end portion of the retention sleeve is turned in over the second abutment part to retain the magnets in position on the carrier sleeve.
3. The motor according to claim 1, wherein the abutment part comprises a ring engaging a shoulder on the carrier sleeve.
4. The motor according to claim 1, wherein the rotor further comprises a plurality of tubular elements supporting the permanent magnets and spaced axially along the shaft.
5. The motor according to claim 1, further comprising:
- a stator coaxial with the rotor and comprising: a stack of laminations; and radially spaced coils wound around the stack.
6. The motor according to claim 1, wherein the motor is an AC synchronous permanent magnet motor.
7. The motor according to claim 1, wherein the motor is capable of reliably operating at speeds greater than 4,500 rpm.
8. The motor according to claim 1, wherein:
- the carrier sleeve is loosely fitted on the shaft, and
- the rotor further comprises support rings closely engaging the shaft and supporting the carrier sleeve.
9. The motor according to claim 8, wherein the carrier sleeve is keyed to the shaft to prevent relative rotation between the carrier sleeve and the shaft.
10. The motor according to claim 9, wherein a key extends outwardly from the shaft and engages complementary locating portions of the carrier sleeve and associated support ring to prevent relative rotation between the carrier sleeve, the support ring and the shaft.
11. The motor according to claim 1, wherein the plurality of permanent magnets is equiangularly spaced about the shaft.
12. The motor according to claim 1, wherein the permanent magnets incorporate axially laminated parts to reduce eddy current losses.
13. The motor according to claim 1, further comprising:
- a bearing mounting the rotor to the stator;
- a resiliently biased projection disposed on one of the stator and the bearing; and
- a receiver disposed on the other of the stator and the bearing,
- wherein the projection is operable, by rotation of the rotor, to engage the receiver, thereby rotationally coupling the bearing and the stator.
14. The motor according to claim 13, wherein the projection is provided on the outer of the stator and the bearing, and the receiver is provided on the inner of the stator and the bearing.
15. The motor according to claim 1, further comprising:
- a housing, wherein the stator is mounted in the housing;
- an axial groove formed in one of the stator and the housing; and
- an axial key engaging the axial groove, thereby rotationally coupling the housing and the stator.
16. The motor according to claim 1, wherein the abutment part comprises a snap ring disposed within a groove in the carrier sleeve.
17. The motor according to claim 1, wherein the retention sleeve is deformed into a detent on the outer surface of the abutment part.
18. An electric submersible pump incorporating a permanent magnet motor according to claim 1.