Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
CA 02614817 2007-11-28
!
~c,D Pt)MP COMOL SYSTBNI INCLUDING PARAMBTSR BSTIMATOR
This application is a divisional application of
Canadian Patent Filel No.2,443,010 filed September 26,
2003.
E&CI0(3ROUND OF THE INYSNTION
[0002] Field of the invention -- The present
invention relates generally to control of rod pumps
for oil and gas wells, and in particular to methods
for optimizing the operation of rod pumps using
parameter estimation.
[0008] The load upon and position of the rods that
drive downhole pumps are important parameters for
control, monitoring, and protection of the artificial
lift system used in oil and gas production. Existing
methods of ineasuring these parameters involve the
mounting and use of external instruments such as
strain gauges, load cells, and position transducers.
The need for these additional devices increases the
1010962
1 -
CA 02614817 2007-11-28
cost and complexity of the pumping system and reduces
system reliability. Generally, AC induction motors
drive rod pumping syatems.
[0004] One method for determining rod load or force
is disclosed in United States Patent No. 4,490,094
(the '094 Patent). With this method, motor velocity
is determined during a complete or predetermined
portion of a reciprocation cycle and the results are
used to compute one or more parameters of pumping unit
performance.
[0005] However, determination of rod load PRLf on
an ith revolution of the prime mover rotor depends on
knowing the position of crank for computation of a
torque factor TFi according to the equation (1):
PRLi=n.Ti+m.sin(Ti+(i) -RIT+AIT (1)
TFi
[0006] Because the torque factor TFi appears in the
denominato'r of the equation, special care must be
taken in deriving the torque factor Tfi and in using
it in the computation to avoid dividing by zero or by
small numbers that would distort the result.
Moreover, the 1094 Patent does not disclose how to
estimate crank position.
[0007] United States Patent No. 5,252,031 (the '031
Patent) discloses a method for monitoring a rod pumped
well to detect various problems. The method uses
measurements made at the surface to calculate a
downhole pump dynamometer card. This downhole pump
dynamometer card is useful in detecting various pump
problems and controlling the pumping unit. The method
involves finding rod position from motor revolutions,
a reference switch and pump geometry. This method
requires setting up look-up tables.
1010962 2-
CA 02614817 2007-11-28
: =
[0008] In addition, the methods disclosed in both
the '031 Patent and the '094 Patent employ a sensor to
detect a rotation of the motor shaft. Because of the
ratio between motor and pump rotations, this method
can produce numerous sample points per stroke of the
pump. However, the time between motor revolutions to
get motor velocity as well as sample other parameters,
such as motor current, is a function of pump speed and
is not suitable for precise monitoriing of the pump
operation. In addition, the method of determining
motor torque relies on a look-up table of steady-state
motor operation rather than a true dynamic calculation
of torque. These methods would work fine for
providing simple pump control function, such as
shutting down the pump when it is pumped off.
However, these methods would not be suitable for real
time closed-loop pump control, such as rod load
limiting, that requires a high bandwidth feedback
signal.
[0009] Past work involving the analysis of rodpump
systems can be divided into two categories. one such
category involves predicting the performance of a rod
pump unit by calculating surface load from known
surface position and assumed pump load. An example of
this method for deriving the surface dynamometer card
from the downhole dynamometer card is disclosed in an
article entitled "Predicting the Behavior of Sucker-
Rod Pumping Systems", by S. G. Gibbs, in JPT, July
1963, pages 769-78, Trans, AIME 228. This uses a
multisection model of=the rod string to simul'ate the
pump operation.
[0010] The other category deals with the diagnosis
of existing pumping installations by determining
actual pump conditions from measured surface
1010962 3-
CA 02614817 2007-11-28
~ conditions. United States Patent No. 3,343,409
discloses a method for =estimating the downhole
dynamometer card from the surface dynamometer card
using frequency based Fourier analysis. However, this
method requires a large number of coefficients to
accurateLy model the high frequency components that
produce the corners of the dynamometer card. In
addition, the method relies on external sensors for
polished rod load and position.
[0011) The average output flow rate of a sucker rod
pump is a function of the, downhole pump stroke and the
average speed of the pump. With existing
technologies, the downhole stroke of the pump is
dictated by the speed of the pumping unit and the
given characteristics of the pumping unit geometry and
the sucker rod stiffness. Significant stretch in the
sucker rod, particularly for deep wells, reduces the
amount of surface rod stroke that can be delivered to
the downhole pump. Additionally, the speed of the
pumping operation is often limited by the need to
avoid overstressing the sucker rod and/or the pumping
unit gearbox. Therefore, output flow rate is
constrained by the imposed pump stroke and stroking
rate.
SUMMARY OF THE INVENTION
[0012] The disadvantages and limitations of the
background art discussed above are overcome by the
present invention. With this invention, 'there is
provided a method of continuously determining
operational parameters of a rod pump used in oil and
gas production, wherein the rod pump includes a rod
string carrying a downhole pump, the rod string
1010962 4-
CA 02614817 2007-11-28
including a polished rod, and a drive system=including
an AC electrical drive motor having a rotor coupled to
the rod string through a transmission unit. Themethod comprises the steps of
continuously measuring
the electrical voltages applied to the drive motor to
produce electrical voltage output signals;
continuously measuring the electrical currents applied
to the drive motor to produce electrical current
output signals; deriving values of instantaneous
electrical torque from the electrical voltage output
signals and the electrical current output signals;
deriving values of instantaneous motor velocity from
the electrical voltage output signals and the
electrical current output signals; and using geometry
of the rod pumping unit and one of the instantaneous
values to calculate instantaneous values of an
operating parameter of the rod pump. In one
embodiment, the method is used for calculating rod
load and/or rod position of a rod pump. The method
also provides calculations of other pump parameters
such as gearbox torque and pump stroke that are useful
in protecting the pumping mechanism and diagnosing
pump problems.
[00131 The invention provides a method of deriving
operating parameters, such as rod load and position,
from the drive motor and pumping unit parameters
without the need for external instrumentation such as
down hole sensors, acoustic fluid level sensors, flow
sensors, etc. The method provides nearly
instantaneous readings of motor velocity and torque
which can be used for both monitoring and real-time,
closed-loop control of the rod pump. In addition,
American Petroleum Institute specification geometry
and system identification_ routines are used to
1010962 5-
CA 02614817 2007-11-28
.
establish parameters used in calculating the
performance parameters that are used in real time
closed loop control of the operation of the rod pump,
obviating the need to create large look-up tables for
parameter values used in calculating performance
parameters. Simple parameters defining the special
geometry used in belt driven pumping units are also
included in the control.
[00141 In one embodiment, wherein the first and
second operating parameters are instantaneous position
and load of the polished rod, the method includes the
steps of using the estimated values of position and
load for the polished rod to obtain a surface
dynamometer card for the rod pump, and deriving from
the surface dynamometer card the instantaneous
position and load of the downhole pump for pump
control and/or generation of a downhole dynamometer
card for the pump.
[0015] The parameter estimator reduces the cost and
complexity of rod pumping systems and provides rod
load measurement accuracy superior to systems using
sensors such as strain gages and load cells.
Moreover, this eliminates wires to sensors mounted on
moving portions of the pump and reliability issues
related to the sensors and their associated wiring.
[0016] Further in accordance with the invention,
the parameter estimator produces values of rod pump
parameters which can be used in optimizing' the
operation of the rod pump. ' Thus, in accordance with a
further aspect of the invention, there are provided
several methods of controlling the rod load and/or
flow rate=of a rod pump used in oil and gas production
and/or preventing damage to the pump assembly, wherein
the rod pump includes a rod string including a
1010962 6-
CA 02614817 2007-11-28
polished rod and a drive system including an AC
electrical motor having a rotor that is coupled
through a transmission unit to the rod string for
reciprocating a downhole pump.
[0017] One method for rod load control uses the
computed rod load to control the force in the rod and
thereby prevent damage to the rod string due to
excessive tension or compression of the sucker rod.
Increased pump speeds will typically produce large
tensile force excursions on the up stroke and large
compressive forces on the downstroke. The method
limits those excursions to preset limits by
manipulating the pumping speed. A second aspect of
the method provides for intentionally increasing- or
decreasing rod load during certain portions of the
pump cycle to increase pump stroke and associated
fluid production.
[0018] Another method of rod pump control provides
for the use of a model of the rod string to derive a
factor for modulating pump speed that reduces rod peak
loads, damps rod force excursions, reduces gearbox
torque loading, increases pump stroke, and improves
energy efficiency without the need for external rod
load and position sensors. Several embodiments of
this method use somewhat different models for control
of the pump. Those models include the use of rod load
andJor rod position to generate control signals that
manipulate pump operation.
[0019] The rod pump control method comprises the
steps of obtaining a measure of the velocity of the
polished rod in real-time; obtaining a measure of
polished rod load in real-time; obtaining an estimate
of the velocity of the pump in real-time; deriving a
modulating factor from the difference between the
1010962 7-
CA 02614817 2007-11-28
velocity of the polished rod and the estimated ~pump
velocity; and using the modulating factor to modulate
motor speed to cause the downhole pump to more closely
follow the polished rod position without excessive
excursions in rod load.
[0020] The invention allows the stroke of the
downhole pump to be increased without an increase in
overall average pumping speed. This increases well
fluid production without increasing overall pumping
speed and enables increased output in wells that are
running at maximum physical capacity of the pumping
system. Alternatively, the method can maintain well
output with decreased overall pumping speed, reduced
rod stress fluctuation, and improved energy
efficiency.
[0021] In accordance with a further aspect of the
invention, there is provided a system for continuously
determining operating parameters of a rod pump used in
oil or gas production, the rod pump including a rod
string carrying a downhole pump driven by an
electrical drive motor that is coupled to the rod
string through a transmission unit. The system
comprises means for determining the torque and
velocity inputs to the rod pump, means for using the
torque and velocity inputs to calculate one or more
values representing the performance of the rod pump,
and means for using parameters related to the geometry
of the rod pump and at least one of said performance
values to calculate values of at least one operating
parameter of the rod pump.
[0022] The rod pump control reduces peak rod loads,
prevents compressive rod forces, and dampens rod load
oscillations thereby reducing rod fatigue and rod
failure. In addition, 'the rod pump control reduces
1010962 8-
CA 02614817 2007-11-28
peak pump velocity, resulting in less power lost to
viscous pump friction, increasing pumping efficiency
and reducing pump wear. Moreover, internal frictional
losses in the rod are reduced by damping rod
oscillations, thereby increasing pumping efficiency.
DESCRIPTION OF THE DRAWINGS
[0023] These and other advantages of the present
invention are best understood with reference.to the
drawings{ in which:
[0024] FIG. 1 is a simplified representation of a
rod pump system including a rod pump control system
that includes a parameter estimator in accordance with
the present invention;
[0025] FIG. 2 is a block diagram'of the rod pump
control system of FIG. 1;
[0026] FIG. 3 is a block diagram of the parameter
estimator of the rod pump control system for
calculating values including gearbox torque, polished
rod load, and rod position using parameters of the
drive-motor and rod pumping unit in accordance with
the present invention;
[0027] FIG. 4 is a block diagram of a process for
obtaining an estimate of rotary weight torque for the
process of FIG. 3;
[0028] FIG. 5 is a block diagram of a process for
obtaining an estimate of total reflected inertia for
the process of FIG. 3;
[0029] FIG. 6 is a block diagram of -a process for
obtaining an estimate of rod load for the process of
FIG. 3;
[0030] FIG. 7 is a block diagram of a process for
selecting rod stroke regions that have torque factors
of sufficient magnitude to produce accurate
1010962 9-
CA 02614817 2007-11-28
=
measurement of rod load as the rod load windowing of
FIG. 3;
[0031] FIG. 8 is a process flow chart for
calculating polished rod load and rod position for the
=5 pump system of FIG. 1, in accordance with the present
invention;
[0032] FIG. 9 illustrates a simulated surface and
downhole dynamometer card for'a conventional beam pump
as well as a downhole dynamometer card generated by
the wave method of computation;
[0033] FIG. 10 illustrates simulated surface and
downhole dynamometer cards for a conventional beam
pump from a commercially available rod pump simulation
program;
[0034] FIG. 11 illustrates a measured surface
dynamometer card for a belt driven pump as well as a
downhole dynamometer card generated by the wave method
of computation;
[0035] FIG. 12 illustrates simulated surface and
downhole dynamometer cards for a bel.t driven pump from
a commercially available rod pump simulation program;
[0036] FIG. 13 is a block diagram of a system for
estimation of rod pump surface and downhole
dynamometer cards for the rod pumping unit in
accordance with the present invention;
[0037] Fig.= 14 is a block diagram of a rod load
control in accordance with the present invention;
[0038] FIG. 15 is a block diagram of a single
section simulation model based rod pump control in
accordance with the present invention;
[0039] FIG. 16 is a block diagram of a multisection
simulation model based rod pump control in accordance
with the present invention;
1010962 10-
CA 02614817 2007-11-28
[0040] FIG. 17 is a block diagram of a wave
equation based rod pump control in accordance with the
present invention;
[0041] FIG. 18 is a process flow chart for
producing rod pump control for improved operation of
the rod pump system of FIG. 1.
[0042] FIG. 19 is a surface dynamometer card for a
beam pump running without rod pump control;
[0043] FIG. 20 is a surface dynamometer card for a
beam pump running with rod pump control;
[0044] FIG. 21 is a downhole dynamometer card for a
beam pump running without rod pump control;
[0045] FIG. 22 is a downhole dynamometer card for a
beam pump running with rod pump control;
[0046] FIG. 23 is a graph showing pump velocity as
a function of time for a beam pump running without rod
pump control;
[0047] FIG. 24 is a graph showing pump velocity as
a function of time for a beam pump running with rod
pump control; and
[0048] FIG. 25 is a block diagram of a processor of
the rod pump control system of FIG. 2.
Definitions of technical terms
The following are definitions of some of the
technical terms used in 'the detailed description of
the preferred embodiments.
Beam Weight (Wb): The equivalent weight of the beam
that is used to calculate its articulating
inertia.
Counterweight Angle (At): The angle of the crank
counterweight offset.
1010962 si-
CA 02614817 2007-11-28
Counterweight Inertia '(Jc): The effective inertia of
the counterweight.
Crank Angle (Ac) The angular position of the
beam pump crankshaft at the output of the
reduction gearbox with respect to a reference
point.
Crank Velocity (Wc): The change in crank angle as a
function of time. The time derivative of the
crank angle.
Downhole Pump Velocity (Vp): The velocity of the
downhole pump as determined by the rod
string/pump simulation algorithm.
Electrical Torque (Te): The torque generated at the
motor shaft as determined from the motor voltages
and currents.
Excitation Frequency (We) : The fundamental frequency
of the instantaneous current circulating in the
drive motor.
Gearbox Output Torque (Tn): The torque at the output
of the gearbox.
Motor Inertia (Jm): The inertia of the motor and
associated components rotating at the motor
speed.
Motor Velocity (Wr): The feedback velocity of the
motor as determined from the motor voltages and
currents.
Overall Gear Ratio (Ng): The gearing reduction
between the motor output shaft and the crank
shaft of the pumping unit. The pumping unit gear
ratio.
Rod Load (Fr): The load applied to the polished rod
as determined by the motor torque, pumping unit
geometry, and pumping system parameters .
Rod Position (Xr): The position of the polished rod
1010962 12-
CA 02614817 2007-11-28
, as determined by the motor position and the
pumping unit geometry.
Rod Velocity (Vr): The velocity of the polished rod
as determined by the motor velocity and the
pumping unit geometry.
Rotary Weight Torque (Tr): The torque component seen
at the gear box output shaft due to the
counterweight normal force.
Torque Command (Tc): The final torque command to the
drive system controlling the pump motor.
Torque Factor (Tf): A factor that, when multiplied by
the load at the polished rod, gives the torque at
the crankshaft of the pumping unit reducer.
Total Reflected Inertia (Jt): The inertia seen at the
motor shaft consisting of motor inertia and
associated high speed components and the
reflected inertias of the counterweight mass and
beam mass.
Unbalanced Force (Bu): The force that would be
required to bring the beam of the pumping unit
to a horizontal position if the unit had no
counterbalance.
Velocity Request (Wx): The pumping unit prime mover
operator requested run speed.
Velocity Command (Wy): The pumping unit prime mover
command velocity. This signal is a conditioned
version of the operator requested run speed, and
originates in the drive control software.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
C00491 Referring to FIG. 1, there is shown a rod
pump system 20, the operation of which is controlled
by a rod pump control system and method including a
parameter -estimator in accordance with the present
1010962 13-
CA 02614817 2007-11-28
invention. For purposes of illustration, the rod pump
control system 21 is described with reference to an
application in a rod pump system 20 that includes a
conventional beam pump. The beam pump has a walking
beam 22 that reciprocates a rod string 24 that
includes a polished rod portion 25. The rod string 24
is suspended from the beam for actuating a downhole
pump 26 that is disposed at the bottom of a well 28.
However, the rod pump control system and method
provided by the invention are applicable to any system
that uses an electric motor to reciprocate a rod
string, including those that drive the rod through
belt or chain drives. For example, a belt driven
pumping unit includes a belt that is coupled to a rod
string for reciprocating the rod string vertically
within a well as the belt is driven by a motor.
[0050) The walking beam 22, in turn, is actuated by
the pitman arm 31 which is reciprocated by a crank arm
30 driven by an electric motor 32 that is coupled to
the crank arm 30 through a gear reduction mechanism,
such as gearbox 34. The typical motor 32 can be a
three-phase AC induction motor operable at 460 VAC and
developing 10-125 horsepower, depending upon the
capacity and depth of the pump. Other types of motors
such as synchronous motors can be used to drive the
pumping unit. The gearbox 34 converts motor torque to
a low speed but high torque output for driving the
crank arm 30. The crank arm 30 is provided with a
counterweight 36 that serves to balance the rod string
24 suspended from the beam 22 in the manner known in
the art. Counterbalance can also be provided by an
air cylinder such as those found on air-balanced
units. Belted pumping units may use a counterweight
1010962 14-
CA 02614817 2007-11-28
= that run in the opposite direction of the rod stroke
or an air cylinder for counterbalance.
[0051] The downhole pump 26 is a reciprocating type
pump having a plunger 38 attached to the end of the
rod string 24 and a pump barrel 40 which is attached
to the end of tubing in the well 28. The plunger 38
includes a traveling valve 42 and a standing valve 44
positioned at the bottom of the barrel 40. On the up
stroke of the pump, the traveling valve 42 closes and
lifts fluid, such as oil and/or water, above the
plunger 38 to the top of the well and the standing
valve 44 opens and allows additional fluid from the
reservoir to flow into the pump barrel 40. On the
down stroke, the traveling valve 42 opens and the
standing valve 44 closes in preparation of the next
cycle. The operation of the pump 26 is controlled so
that the fluid level maintained in the pump barrel 40
is sufficient to maintain the lower end of the rod
string 24 in the fluid over its entire atroke.
[0052] Referring to FIG. 2, which is a simplified
representation of the rod pump control system 21
including parameter estimator in accordance with the
present invention, the parameter estimator determines
parameters relating to operation of the rod pump from
motor data without the need for external
instrumentation. In one embodiment, instantaneous
motor currents and voltages together with pump
parameters are used in determining rod position and
load without the need for strain gauges, load cells,
or position sensors as well as determining pump
pressure and pump flow without the need for additional
downhole or surface sensors. The rod position and
load can be used to control the operation of the pump
26 to =optimize the operation of the pump 26. In
1010962 15-
CA 02614817 2007-11-28
addition, American Petroleum Institute (API)
specifications have been used to define the pump
geometry that allows the use of readily available data
from pump manufacturers. System identification
routines are used to establish installation dependent
parameters specific to the =particular pump used in
calculating performance parameters that are used in
real-time closed loop control of the operation of the
rod pump, obviating the need to create large look-up
tables for parameter values used in calculating
performance parameters.
[0053] The pump control system 21 includes
transducers, such as current and voltage sensors, to
sense dynamic variables associated with motor torque
and velocity. As shown in FIG. 2, current sensors 50
are coupled to a sufficient number of the motor leads
for the type of motor used. The current sensors 50
provide voltages proportional to the instantaneous
stator currents in the motor 32. Voltage sensors 52
are connected across to a sufficient number of the
motor windings for the type of motor used and provide
voltages proportional to the instantaneous voltages
across the motor windings. The current and voltage
signals produced by sensors 50 and 52 are supplied to
a processor 53 through suitable input/output devices
54. The processor 53 further includes a processing
unit 55 and storage devices 56 which stores programs
and data files used in calculating operating
parameters and producing control signals for
controlling the operation of the rod pump system 20.
This control arrangement provides nearly instantaneous.
readings of motor= velocity and torque which can be
used for both monitoring and real-time, closed-loop
control of the rod pump. For example, in one
1010962 16-
CA 02614817 2007-11-28
embodiment, computations of motor velocity and torque
used for real-time, closed-loop control are provided
at the rate of 1000 times per second.
100541 Motor currents and voltages are sensed to
determine the instantaneous electric power level drawn
from the power source by the electric motor operating
the well pump. As the rod string 24 that drives the
downhole pump 26 is raised and lowered during each
cycle, the motor 32 is cyclically loaded. Depending
on the particular pump installation configuration, the
walking beam 22 is at a known position during maximum
and minimum motor loads. The timing of these maximums
and minimums can define the operational pumping
frequency and, by integration of the motor velocity in
light of the motor to crank gearing, it is possible to
estimate the phase position of the pump crank at any
time. By monitoring the variances of the motor
currents and voltages as a function of pump crank
angle, the voltage and current variances can be used
together with parameters related to pump geometry to
calculate estimates of rod position Xr and rod load
Fr.
(0055] Referring to FIG. 3, there is shown a block
diagram of a parameter estimator 23 of the rod pump
control system 21 for calculating estimates of
parameters including rod position Xr, rod load Fr, and
gearbox output torque Tn. In one preferred
embodiment, the calculation is carried out by the
processing unit 55 (FIG. 2) under the control of
software routines stored in the storage devices 56.
Block 62 responds to signals corresponding to
instantaneous values of motor current and voltage to
produce a measure of electrical torque Te of the drive
motor 32. Block 63 responds to the signals
1010962 17-
CA 02614817 2007-11-28
corresponding to instantaneoue values of motor current
= and voltage to produce an estimate of velocity Wr of
the drive motor 32. Block 64 calculates rod position
Xr and torque factor Tf. - Block 65 calculates an
estimate of rod load Fr. Block 67 calculates an
estimate of rotary weight torque Tr. Block 68
calculates total reflected inertia Jt. Block 69
produces an output corresponding to acceleration Alpha
of the drive motor shaft.
[0056] More specifically, blocks 62 and 63 can
include hardware circuits which convert and calibrate
the motor current and voltage signals provided by the
sensora or transducers 50 and 52 (FIG. 2) into current
and flux signals. The hardware circuits scale and
translate the current and flux signals into an
internal frame of reference. After scaling and
translation, the outputs of the voltage and - current
sensors can be digitized by an analog to digital
converter. Block 62 combines the scaled signals with
motor equivalent circuit parameters to produce a
precise measure of electrical torque Te. Automatic
identification routings can be used to establish the
motor, equivalent circuit parameters. Block 63
combines the scaled signals with motor equivalent
circuit parameters to produce a precise measure of
motor velocity Wr.
[0057] In one embodiment, the stator flux is
calculated from motor voltages and currents -and the
electromagnetic torque is directly estimated from the
stator flux and stator current. Three-phase motor'
voltages and currents are converted to dq
(direct/quadrature) frame signals using three to two
phase conversion for ease of computation in a manner
known in the art. Signals in the dq frame can be
1010962 18 -
CA 02614817 2007-11-28
represented as individual signals or as vectors for
convenience. Block 62 responds to motor stator
voltage vector Vs and motor stator current vector Is
to calculate a measure of electrical torque Te
produced by the motor. In one embodiment, the
operations carried out by block 62 for calculating the
electrical torque estimate are as follows. The stator
flux vector Fe is obtained from the motor stator
voltage Vs and motor stator current Is vectors
according to equation (2):
Fs= (Vs-Is.Rs) /s (2)
Fds=(Vds-Ids.Rs)/s (2A)
Fqs= (Vqs-Iqs.Rs) /s (2B)
where Rs is the stator resistance and s(in the
denominator) is the Laplace operator for
differentiation. Equation (2A) and (2B) show typical
examples of the relationship between the vector
notation for flux Fs' , voltage Va, and current Is and
actual d axis and q axis signals.
[0058] In one embodiment, the electrical torque me
is estimated directly from the stator flux vector Fs
obtained from equation (2) and the measured stator
current vector Is according to equation (3) or its
equivalent (3A) :
Te=Ku. (3/2) .P.FsxIs (3)
Te-Ku. (3/2) .P. (Fds.Iqs-Fqs.Ids) (3A)
where P is the number of motor pole pairs and Ku is a
unit scale factor to get from MKS units to desired
units.
[0059] In one embodiment, rotor velocity Wr is
obtained from estimates of electrical frequency We and
1010962 19-
CA 02614817 2007-11-28
slip frequency Ws. The inputs to block 63 also are
the stator voltage Vs and stator current Is vectors.
Block 63 calculates the motor velocity Wr. In one
embodiment, the operations carried out by block 63 for
calculating the motor velocity are as follows. A
rotor flux vector Fr is obtained from the measured
stator voltage Vs and stator =current Is vectors along
with motor stator resistance Re, stator inductance Ls,
magnetizing inductance Lm, leakage inductance SigmaLs,
and rotor inductance Lr according to equations (4) and
(5); separate d axis and q axis rotor flux
calculations are shown in equations (5A) and (SB)
respectively:
SigmaLs=Ls-LmA2/Lr (4)
then,
Fr= (Lr/Lm). [Fs-Is.SigmaLs] (5)
Fdr= (Lr/Lm).(Fds-SigmaLs.Ids) (5A)
Fqr= (Lr/Lm).(Fqs-SigmaLs.Iqs) (5B)
[0060] The slip frequency Ws can be derived from
the rotor flux vector Fr, the stator current vector
ie, magnetizing inductance Lm, rotor inductance Lr,
and rotor resistance Rr according to equation (6):
Ws=Rr.(Lm/Lr).[Fdr.Iqs-Fqr.Ids]
FdrA2+FqrA2 (6)
[0061] The instantaneous excitation or electrical
frequency We can be derived from stator flux
according to equation (7):
We = Fds.sFqs-Fqs.sFds
FdsA2+FqsA 2 ( 7 )
1010962 20-
CA 02614817 2007-11-28
[0062] The rotor velocity or motor velocity Wr can
be derived from the slip frequency Ws and the
electrical frequency We according to equation (8):
Wr= We-Ws (8)
[0063] The motor velocity Wr is passed through an
amplifier 70 and divided by the. gain Ng which
represents the overall gear ratio between the motor
and the pump crankshaft. Consequently, the motor
velocity Wr that has been obtained from motor voltage
and current is converted to crank velocity Wc, which
reflects the overall pumping unit gear ratio, that is
being produced at the output of the gearbox 34.
[0064] The crank velocity Wc is integrated in block
71 to obtain a position which, when combined with the
overall pumping unit gear and a reference position,
yields the angular position that is the crank angle Ac
~
of the pumping unit gearbox. The reference position
can be obtained using a: magnetic or optical sensing
device, a cam limit switch, or similar device, to
define a reference point within the stroke of the
pumping unit for each cycle of operation.
[0065] Block 64 calculates the rod position Xr and
the torque factor Tf using the crank angle Ac obtained
from the crank velocity Wc, and parameters associated
with beam pump 'geometry. As is known, the geometry
of the pumping unit is defined by the American
Petroleum Institute and can be entered directly into
the control in that format. One source of API
specifications is API Specification 11E, entitled
"Specification for Pumping Units , seventeenth
edition, November 1, 1994. Information entered is
1010962 21-
CA 02614817 2007-11-28
dependent upon the class of the rod pump and direction
of rotation. Typical beam pump parameters that are
used for calculating the rod position Xr include the
dimensions of the walking beam, crank radius, and
pitman arm as well as the location of the various
pivot points in the unit. Those pump parameters are
readily available from pumping unit manufacturers.
Simple parameters are also included in the control for
belt type pump mechanisms that are not specified by 10 the API standard.
Automatic identification routings
are used to establish installation dependent pumping
unit parameters such as counterbalance inertia and
frictional terms.
[00661 Block 67 combines the crank angle Ac with
the counterweight angle At to produce an estimate of
rotary torque Tr associated with the weight of the
counterweight. Referring to FIG. 4, in one
embodiment, the rotary torque Tr is bbtained by
summing the crank angle Ac with the counterweight
angle At using summing block 72. Block 73 obtains the
sine of the resultant value. The result is passed
through amplifier 74, the gain -Mu of which is
selected to correspond to the counterweight moment,
producing the rotary torque Tr.
[0067] Referring to FIGS. 3 and 5, block 68
combines torque factor Tf, produced by block 64, beam
weight Wb, counterweight inertia Jc and motor inertia
Jm to produce total reflected inertia Jt. With
reference to FIG. 5, block 75 obtains the product of
torque factor Tf and equivalent beam weight Wb.
Torque factor Tf is entered twice to square that
factor. In one embodiment, the gain of amplifier 76
is ijG which divides the product by the acceleration
of gravity such that output of amplifier 76 is the
1010962 22-
CA 02614817 2007-11-28
.
. articulating inertia of the pumping unit. The result
is combined with counterweight inertia Jc at the crank
in summing block 77 and scaled by amplifier 78, the
gain 1/Ng"2 of which is selected to correspond to the
inverse square of the overall gear ratio. The scaled
value is combined with motor inertia Jm by summing
block 79 to obtain the estimate of total reflected
inertia Jt.
[0068] Referring to FIGS. 3 and 6, block 65
combines electrical torque Te calculated by block 62,
rotary weight torque estimate Tr calculated by block
67, an estimate of total reflected inertia Jt
calculated by block 68, motor acceleration estimate
Alpha, produced by block 69, torque factor Tf from
block 67, static friction Sf, crank velocity Wc from
amplifier 70, viscous friction factor Bf, and
unbalanced force Bu to produce the rod load estimate
Fr. With reference to FIG. 6, block 80 obtains the
product of motor acceleration estimate Alpha and an
estimate of total reflected inertia Jt at the motor,
the result of which is subtracted from the electrical
torque Te by summing block 81. This difference is
scaled by a factor corresponding to the gearbox ratio
Ng, using amplifier 82. The difference between
electrical torque as modified by motor acceleration
and load inertia factor, provided by amplifier 82
minus the static torque Sf, provided by setup testing,
and viscous torque, provided by multiplying in
amplifier 84 the crank speed Wc by a viscous friction
factor Bf determined,during setup, and rotary weight
torque Tr is divided by the torque factor Tf in block
85, and the result is summed with unbalanced force Bu
in summing block 86 to produce rod load estimate Fr.
A rod load update enable switch 87 and memory element
1010962 23-
CA 02614817 2007-11-28
- 88 are used to hold the prior value of rod load at
.
around the 'points where the torque factor Tf goes to
,zero as determined by, the rod load update enable
output of block 60 detailed in FIG 7.
[0069] Referring to FIGS. 3 and 7, block 60
compares the rod position to the positions defined by
the ends of the rod stroke Sr determined during setup.
If the rod position is within the deadzone Dz of
either end of the stroke, the rod load update enable
output is off and block 65 (FIG. 6) is inhibited from
updating the rod load Fr. Deadzone Dz is also
determined during setup. With reference to FIG. 7,
relational operator 89 compares the current value of
rod position Xr to the rod load deadzone value Dz and
outputs a logical true if Xr is greater than Dz.
Summing block 90 subtracts the rod load deadzone value
Dz from the rod stroke Sr. Relational -operator 91
compares the current value of rod positioin Xr to the
output of summing block 90 and outputs a logical true
if Xr is the lesser value. Logical operator 92
outputs a logical true only if both relational
operators 89 and 91 are outputting logical trues.
[0070] Referring again to FIG. 3, the gearbox
output torque Tn can be computed from the eleetrical
torque Te produced by block 62 by using the overall
gearbox ratio. The value of electrical torque Te
produced by block 62 is multiplied by a factor related
to gearbox ratio Ng, using an amplifier 66, to provide
an estimate of gearbox output torque Tn.
[0071] Block 69 produces an output corresponding to
acceleration Alpha of the drive motor shaft. One
method to obtain motor acceleration in Alpha Block 69
is to differentiate the motor velocity Wr.
1010962 2 4 -'
CA 02614817 2007-11-28
=
[0072] Multiplier block 61 produces an output
corresponding to rod velocity Vr by computing the
product of torque factor Tf and crank velocity Wc.
[0073] Referring to FIG. 8, there is shown a
process flow diagram for obtaining estimates of
polished rod position Xr, polished rod load Fr, and
gearbox torque Tn derived from the motor current and
voltage in accordance with the invention. At start-
up, automatic identification routines are used offline
to estimate various parameters. In one embodiment,
the automatic identification routines determine
overall gear ratio Ng and counterweight moment Mu for
use in further calculations. The overall gear ratio
is the difference between the motor revolutions and
the crank cycle. The automatic identification
routines also are used to establish motor equivalent
circuit parameters as well as installation dependent
pumping unit parameters, such as static friction
torque Sf and viscous friction factor Bf.
[0074] Referring aleo to FIG. 3, after
initialization, block 93 obtains the instantaneous
values of motor stator current and motor stator
voltage from sensors 50 and 52, respectively. As
described above, blocks 62 and 63 respond to the motor
current and voltage signals from the sensors 50 and 52
for use in calculating motor torque and velocity.
Motor stator ciurrent and voltage are measured
continuously allowing the instantaneous values of
current and voltage to be obtained through the
measurement.
[0075] In block 94, the values of instantaneous
motor current and motor voltage obtained from the
measurements are used to derive electrical torque Te.
In one embodiment, the stator flux is derived from the
1010962 2 5 -
CA 02614817 2007-11-28
motor currents and voltages, using equation (2) as
described above. The electrical torque Te can be
directly estimated from this stator flux-and the motor
current measured, using equation (3).
[0076] In block 95, the values of instantaneous
motor current and motor voltage obtained from the
measurements are used to derive motor velocity Wr. In
one embodiment, rotor flux is obtained from the
measured voltage and current, and stator resistance
and inductance, using equations (4) and (5) -as has
been described. Then, slip frequency is derived from
the rotor flux, the measured motor current,
magnetizing inductance, rotor inductance, and rotor
resistance using equation (6). An estimate of
electrical frequency' is derived from the stator flux
using equation (7) as described above. Then, motor
velocity Wr is derived from the slip frequency and the
electrical frequency using equation (8) as described
above.
[0077] The motor velocity Wr obtained in block 95
is used to obtain crank velocity Wc in block 96. In
one embodiment, the crank velocity is obtained by
scaling the motor velocity as described above with
reference to FIG. 3.
[0078] In block 98, the crank angle Ac is obtained
by integrating the crank velocity Wc obtained in block
96. A limit switch or similar device may be used to
determine a reference point within the stroke of the
pumping unit. The crank velocity Wc is integrated to
get position that combined with the overall pumping
unit gear ratio and reference position give the crank
angle Ac.
[0079] In block 100, rod position Xr is calculated
using, the crank angle Ac together with parameters
1010962 2 6 -
CA 02614817 2007-11-28
associated with pumping unit geometry as described
above with reference to FIG. 3. The use of the system
parameters with crank angle Ac allows the calculation
of rod position Xr.
[0080] In block 102, the gearbox torque Tn is
calculated using the electrical torque Te obtained
from block 92. The overall gear ratio Ng is also used
to compute gearbox output torque Tn from motor
electrical torque Te.
[0081] In block 104, the rotary weight torque Tr,
is calculated by block 67, the total reflected inertia
Jt, is calculated by block 68, and motor acceleration
Alpha is calculated by block 69.
[0082) In block 106, the combination of the system
parameters and pumping unit geometry with electrical
torque Te provides the computation of rod load Fr.
The electrical torque estimate Te is used to obtain
the rod load estimate Fr.
[0083] The method of estimating the load and
position of the polished rod at the surface is
possible without requiring down hole sensors, acoustic
fluid level sensors, flow sensors, etc... The values
of polished rod load and position can be commonly
plotted in XY format to produce a surface dynamometer
card. The estimation method is a real-time,
continuously updating method, i.e,, it is not
performed off-line in a batch manner. Moreover, the
method of estimating a surface dynamometer card for a
rod pump unit does not employ any load or position
transducers.
[0084] In accordance with a further aspect of the
invention, the values of polished rod load and
position can be used to produce a downhole dynamometer
card estimate without the need for sensors. Referring
1010962 27-
CA 02614817 2007-11-28
to FIG. 13, there is shown a block diagram of a system
107 for obtaining a downhole dynamometer card without
requiring down hole sensors, acoustic fluid level
sensors, flow sensors, etc., using the parameter
estimator 23 described above. The system 107 includes
a downhole dynamometer card estimator, block 108, that
uses the polished rod position and the polished rod
load parameter values obtained by the =parameter
estimator 23 to produce an estimation of the downhole
dynamometer card. Thus, the downhole dynamometer card
estimation is produced without the need for rod
position or load transducers. The motor voltage and
current output signals obtained from measurements of
the motor voltage and current are used to derive
instantaneous values polished rod load and position
which are used in producing the estimated downhole
dynamometer card.
[0085] The accuracy of the estimation of the
downhole pump is dependent upon simulating damping
forces that are=inherent in sucker rod pump systems.
A viscous damping coefficient is used to model these
damping forces.
[0086] More specifically, in one embodiment, an
estimation of the downhole dynamometer card is
obtained using the wave equation to model the force
trajectory along the rod string in distance and time.
The wave equation is a linear hyperbolic differential
equation that describes the longitudinal vibrations of
a long slender rod. Using the wave equation with
viscous damping, the motion of a sucker'rod string can
be approximated. The wave equation is used only to
model the rod string and force travelling through it.
The pump sets the boundary conditions for the wave
equation at the bottom and the surface prime mover
1010962 2 8 -
CA 02614817 2007-11-28
. sets the boundary conditions for the wave equation at
the top. The continuous form of the wave equation
with constant rod diameter is:
~u a.2u u
vZ - + c - (9)
3xa 2it2 1 t
where u is the rod displacement, x is the axial
distance along the length of the rod, c is the damping
coefficient, and v is the velocity of force
propagation in the rods.
[0087] Details of the use of the wave equation in
estimating a downhole dynamometer card are disclosed,
for example, in a paper entitled "An Improved Finite-
Difference Calculation of Downhole Dynamometer Cards
for Sucker-Rod Pumps", by T. A. Everitt and J. W
Jennings, SPE 18189, SPE Production Engineering, Feb.
1992, pages 121-127. For simplicity, Equation (9) is
for the case of a constant rod diameter. However, as
disclosed in the referenced paper of T. A. Everitt and
J. W Jennings, with modification, 'this method can also
account for variable rod diameter, including tapered
rod-strings and rod strings of variable density, e.g.,
steel or fiberglass. Solving the wave equation
requires only two boundary conditions because only
steady state solutions are needed. The typical use of
the wave equation would be to use sampled data of a
surface dynamometer card from a rod pumping systems to
do an off-line calculation of the pump downhole
dynamometer card. In this invention, the wave
equation -is solved on-line for each data point so the
results can be used in the next sample period for
control of the pumping system. The two boundary
1010962 29-
CA 02614817 2007-11-28
conditions are polished rod load Fr and position Xr as
a function of time. These conditions are produced by
the parameter estimator 23.
[0088] The damping coefficient c can be similar to
that presented by T. A. Everitt and J. W Jennings in
the referenced paper, or that presented in United
States Patent No. 3,343,409 issued to S. G. Gibbs.
[0089] The accuracy of the downhole dynamometer
card estimate can be verified by performing
simulations. One verification procedure that 'can be
used is similar to that disclosed in the paper by
Everitt and Jennings referenced above.
[0090]' Using the multisection simulation disclosed
in the paper by S. G. Gibbs, referenced above, the
surface dynamometer card load is estimated from a
given surface position trajectory and pump load and
position. This method computes new rod position
estimates in time. Then, using the finite difference
method disclosed by Everitt and Jennings in the paper
referenced above, the downhole dynamometer card is
estimated from the surface dynamometer card generated
previously. Then, the estimated downhole dynamometer
card is plotted with the predicted downhole
dynamometer card to verify the accuracy of the
estimated downhole dynamometer card.
[0091] FIG. 9 demonstrates the ability of the wave
equation method to extract downhole pump operation
from surface information. An assumed full pump
condition indicated by reference number 9-1 was used
to simulate surface dynamometer card parameters 9-3.
Those parameters were ueed with the wave equation
method to generate the predicted downhole pump
dynamometer card 9-2 which closely tracks the
originally assumed pump dynamometer card. FIG. 10
1010962 30-
CA 02614817 2007-11-28
shows the results that were obtained using a
commercially available simulation program to check the
results of the multisection simulation and wave
equation shown in FIG. 9
[0092] The foregoing simulations were conducted for
a conventional beam type rod pump. However, the
finite difference method can be used for estimating
the downhole dynamometer card for other types of rod
pump units, such as a rod pump unit in which the
driver includes a belt drive. FIG. 11 shows results
for a pumping unit including a belt that is coupled to
a rod string for reciprocating the rod string
vertically within awell as the belt is driven by a
motor. The graph given by FIG. 11 includes a surface
dynamometer card 11-1 obtained from actual field
measure data and a predicted downhole dynamometer card
11-2 obtained from the wave equation method. The
source data was captured at a constant pump velocity
of four strokes per minute.
[0093] FIG. 12 illustrates results which are
similar to those illustrated in FIGS. 11 which were
obtained using a commercially available simulation
program.
[0094] The rod load Fr and/or rod position Xr
parameters obtained using the parameter estimator can
be used to provide various control functions. By way
of example, control algorithms can use the rod load,
rod position, or both to achieve improved pump
operation.
[0095] Referring to FIG. 14, there is shown a block
diagram of a system 130 for controlling a pump using
rod load control. This control algorithm uses the rod
load Fr, which can be from the estimator in FIG. 3,
along with maximum upper load and minimum lower load
1010962 31-
CA 02614817 2007-11-28
. parameters to achieve desirable rod loading. Rod
loads can be increased in areas of the pump cycle with
low rod stress to increase pump stroke and associated
production and/or reduced in areas of the pump cycle
with high rod stress to avoid rod damage.
[0096] When the torque factor Tf, which can be from
the estimator in FIG. 3, is positive, the switch 135
causes the upper portion of the control to be
selected. Summing block 131 subtracts rod load Fr from
the value Max Upper_Load, which is determined during
setup, and outputs the result as Fue. If Fue is
greater than zero, switch 133 causes it to be
multiplied by the above upper gain Kau in gain block
136. If Fue is less than or equal to zero, switch 133
causes it to be multiplied by the below upper gain Kbu
in gain block 137.
[0097] Similarly, When the torque factor Tf is zero
or negative, the switch 135 causes the lower portion
of the control to be selected. Summing block 132
subtracts the value Min Lower Load, which is
determined during setup from rod load Fr, and outputs
the result as Fle. If Fle is greater than zero,
switch 134 causes it to be multiplied by the above
lower gain Kal in gain block 138. If Fle is less than
or equal to zero, switch 134 causes it to be
multiplied by the below lower gain Kbl in gain block
138.
[0098] Whichever value is calculated is then
multipled with the absolute value of torque factor Tf
by multiplier block 141. The absolute value of Tf is
derived by the abeolute value block 140. The output
of the multiplier block is added to the velocity
request Wx by summing block 142 to generate the
velocity command Wy.
1010962 32-
CA 02614817 2007-11-28
[00991 Referring to FIG. 15, there is shown the
block diagram of a system 110 wherein the rod load Fr
parameter, which can be from the estimator in FIG. 3,
can be used along with a one section model of the rod
string based on rod stiffness to provide a control
function referred to hereinafter as a rod load damping
control. This control dampens the stress excursions
in the rod string and causes the downhole pump motion
to more closely follow the motion of the polished rod
at -the surface. Therefore, efficiency and reliability
of the pump system is increased.
[0100] Rod load Fr is divided by Rod_Stiffness,
which is determined during setup, in division block
111. The result is differentiated by derivative
function block 112 producing a velocity error term.
If the torque factor Tf, which can be. from the
estimator in FIG. 3, is greater than zero, switch
block 117 causes the velocity error to be multiplied
by the gain factor Kup in gain block 113 and then
multiplied by the torque factor Tf in multiplier block
114. If the torque factor Tf is less than or equal to
zero, switch block 117 causes.the velocity error to be
multiplied by the gain factor Kdn in gain block 115
and then multiplied by the torque factor Tf in
multiplier block 116. The result is then added to the
velocity request Wx by summing block 118 to generate
the velocity command Wy.
[0101] Referring to FIG. 16, there is shown the
block diagram of a system 160, wherein the rod
position Xr and rod load Fr parameters, which can be
from the estimator in FIG. 3, can be used along with a
multisection simulation model of the rod string to
provide a control function referred to hereinafter as
a.simulation model control.
1010962 3 3 -
CA 02614817 2007-11-28
[0102] Rod load Fr and rod position Xr are input to
rod string model block 161. The rod string model
simulates the rod behavior by dividing the rod string
into a finite number of elements. Each element has a
mass and spring constant. The dynamic effects of the
changing rod load Fr and rod position Xr are
calculated on each section to determine the velocity
of the downhole pump.
[0103] The rod velocity Vr, which can be from the
estimator in FIG. 3, is subtracted ,from the pump
velocity in summing block 162 to determine the
velocity error term. If the torque factor Tf., which
can be from the estimator in FIG. 3, is greater than
zero, switch block 167 causes the velocity error to be
multiplied by the gain factor Kup in gain block 163
and then multiplied by the torque factor Tf in
multiplier block 164. If the torque factor Tf is less
than or equal to zero, switch block 167 causes the
velocity error to be multiplied by the gain factor Kdn
in gain block 165 and then multiplied by the torque
factor Tf in multiplier block 166. The result is then
added to the velocity request Wx by summing block 168
to generate the velocity command Wy.
[0104] Referring to FIG. 17, there is shown the
block diagram of a system 170, wherein the rod
position Xr and rod load Fr parameters, which can be
from the estimator in FIG. 3, can be used along with a
wave equation model of the rod string to provide a
control function referred to hereinafter as a wave
equation control.
[0105] The wave equation control is a control
algorithm capable of damping rod load oscillations,
reducing rod stress, and increasing pump stroke
without changing the overall pumping speed, or in the
1010962 3 4 -
CA 02614817 2007-11-28
alternative, maintaining the well output with
decreased overall pumping speed. The wave equation
control according to the invention increases the pump
stroke, decreases peaks. in rod load and dampens rod
load oscillations. However, average pumping speed is
not affected. The wave equation control enables
increased output in wells that are running at maximum
conventional capability of the pumping system.
[0106] The wave equation control manipulates motor
velocity to maximize downhole pump stroke. The
control function provided by the wave equation control
basically consists of estimating pump velocity state
by means of a discrete rod string, fluid, and pump
model. The pump velocity state is then multiplied by
a damping gain and summed with the request velocity.
This lowers the rod load overshoot through active
damping while also increasing the downhole pump
stroke. This results in an increase in output flow
rate without an increase in overall average pumping
speed which, in turn, increases well output without
increasing overall pumping speed. This can provide
increased output in wells that are running at maximum
capacity. Alternatively, a given well output can be
maintained with decreased overall pumping speed.
[01073 More specifically, with reference to FIG.
17, the wave equation control includes a rod string
model 171 in which the rod load Fr and rod position Xr
can be from the parameter estimator of FIG. 3.
[0108] The wave equation control 170 employs a rod
string model (i.e., rod string model 171) that
produces pump velocity Vp and pump position Xp states.
However in one embodiment, only the pump velocity Vp
is used in the control function. Although pump
position Xp is not used for control, pump position can
1010962 35-
CA 02614817 2007-11-28
be used to estimate pump stroke Sp. The pump stroke
information, in turn, can be used to generate flow
rate information.
[0109] The rod/pump simulation 171 responds to rod
position Xr and rod load Fr and produces an output
representative of simulated pump velocity Vp.
[0110] The rod velocity Vr, which can be from the
estimator in FIG. 3, is subtracted from the pump
velocity in summing block 172 to determine the
velocity error term. If the torque factor Tf, which
.can be from the estimator in FIG. 3, is greater than
zero, switch block 177 causes the velocity error to be
multiplied by the gain factor Kup in gain block 173
and then multiplied by the torque factor Tf in
multiplier block 174. If the torque factor Tf is less
than or equal to zero, switch block 177 causes the
velocity error to be multiplied by the gain factor Kdn
in gain block 175 and then multiplied by the torque
factor Tf in multiplier block 176. The result is then
added to the velocity request Wx by summing block 178
to generate the velocity command Wy.
[0111] Referring to FIGS. 18, there is shown a
process flow diagram for producing simulation model
control and wave equation control in accordance with
the invention. Block 150 obtains the polished rod
position Xr. This can be done using the algorithm as
described above with reference to FIG. 3.
[0112] Block 152 obtains the polished rod velocity
Vr. This can be done using the algorithm as described
above with reference to FIG. 3.
[01131 The downhole pump velocity Vp is obtained in
block 154. This is obtained using the rod string
model 161 for the simulation model control or 171 for
the wave equation control.
1010962 3 6 -
CA 02614817 2007-11-28
[0114] Then, the difference of the surface rod
velocity Vr and the downhole pump velocity Vp is
obtained in block 156 by subtracting the pump velocity
from the polished rod velocity, as shown by summing
blocks 162 and 172.
[01151 The modulating factor is created in block
158 by applying the damping difference between the
surface rod velocity and the pump velocity to the
proportional gain amplifiers selected from 163, 165,
173 and 175 by switch blocks 167 and 177 and then
multiplying by the torque factor Tf in blocks 164,
166, 174 and 176.
[0116] The modulating factor is combined with the
velocity request Wx by summing blocks 168 and 178 to
produce a command velocity Wy for the drive motor 32.
The velocity command Wy signal varies 'as a function of
the change in rod velocity Vr relative to pump
velocity Vp.
101171 FIGS. 19-24 illustrate results of a beam
pump running with and without the simulation model
control algorithm enabled. Referring initially to
FIGS. 19 and 20, there is shown a surface dynamometer
card for a pump running without simulation model
control and with simulation model control,
respectively. The data is that fQr a beam pump
running at seven strokes per minute.
[0118] As can be seen by comparing the dynamometer
card in FIG. 19 with the dynamometer card shown in
FIG. 20, with the simulation model control enabled,
the rod stress fluctuation is reduced by lowering the
peak rod up stroke load while the raising minimum rod
down stroke load. For example, the dynamometer card
in FIG 19 shows a peak rod load of about 36,000 pounds
while the dynamometer card in FIG 20 shows a peak rod
1010962 3 7 -
CA 02614817 2007-11-28
load of about 33,000 pounds. In addition, the
dynamometer card in FIG 19 shows a minimum rod load of
about 13, 000 pounds while the dynamometer card in FIG
20 shows a minimum rod load of about 16,000 pounds.
Rod load oscillation is dampened as can be seen by
comparing FIG. 19 with FIG. 20. . Rod load fluctuation
of 17,000 (33,000-16,000) pounds with rod pump control
is 26% less then the 23000 (36,000-13,000) pounds
without simulation model control.
[0119] FIGS. 21 and 22 show the downhole pump
dynamometer cards associated with FIG. 19 and FIG. 20
respectively. As can be seen by comparing the
dynamometer card in FIG. 21, without simulation model
control, with the dynamometer ca'rd shown in FIG. 22,
with the simulation model control, the pump stroke has
increased from 255 inches to 282 inches. This 27 inch
difference translates to an increase in fluid'
production of nearly 11%.
[0120] Additional advantages of simulation model
control can be seen by comparing the graphs in FIGS.
23 and 24. Graphs in those figures show motor
velocity Wr, pump velocity Vp, and rod velocity Vr.
Without simulation model control, FIG. 23, the pump
velocity reaches peak values that are nearly twice
that of the polished rod and considerable time is
spent dwelling at zero velocity. When simulation
model control is enabled, FIG. 24, the peak pump
velocity more nearly tracks polished rod velocity and
no time is wasted dwelling at zero velocity. This
provides for increased pump stroke without the need
for high pump peak speeds.
[0121] In this example, pump stroke is increased
approximately 11t with no overall change in average
pumping unit speed. In addition, peak rod load is
1010962 38-
CA 02614817 2007-11-28
reduced, minimum rod load is increased, rod load
oscillation is dampened, and peak pump velocity is
reduced.
[0122] Referring to FIG. 25, in one preferred
embodiment, the system provided by the present
invention, is software based and is capable of being
executed in a processor 53 shown in block diagram form
in FIG. 25. In one embodiment, the computer system
includes input devices 181, such as current and
voltage sensors connected to analog to digital
converters, output devices 182, such as, a variable
frequency drive, and a processing unit 55 having
associated random access memory (RAM) and read-only
memory (ROM). In one embodiment, the storage devices
include a database 185 and software programs and files
which are used in carrying out simulations of circuits
and/or systems in accordance with the invention. The
programs and files of the computer system include an
operating system 186, the parameter estimation engine
187, and a control method 188 such as the simulation
model control engine, rod load control engine, rod
load damping control engine or wave equation control
engine, for example. The programs and files of the
computer system can also include or provide storage
for data. The processor is connected through suitable
input/output interfaces and internal peripheral
interfaces (not shown) to the input devices, the
output devices, the storage devices, etc., as is
known.
[0123] Although an exemplary embodiment of the
present invention has been shown and described with
reference to particular embodiments and applications
thereof, it will be apparent to those 'having ordinary
skill in the art that a number of changes,
1010962 39-
~-.-...~.~.. -..,._..,T--.~ _.
CA 02614817 2007-11-28
-
modifications, or alterations to the invention as
described herein may be made, none of which depart
from the spirit or scope of the present invention.
All such changes, modifications, and alterations
should therefore be seen as being within the scope of
the present invention.
1010962 4 0 -
