__Background of the Invention__
__Field of the Invention__
This invention relates to a pumping unit according to the preamble
of claim 1 (see also EP-A- 0 297 515) which is adapted for a method and apparatus
for tightening threaded fasteners using a hydraulic torque wrench based on determinations
of parameters representative of torque and angle of a threaded fastener.

__Discussion of the Prior Art__
Threaded fasteners (hereinafter referred to as 'fasteners'), such
as a bolt and nut, a bolt threaded into a bore, or a nut threaded onto a stud or
shank, are commonly used to connect two or more members into a solid rigid structure
or joint. It is highly desirable that the components of the rigid structure remain
in the tightened state at all times, and especially when external loadings such
as vibration, shock and static or dynamic forces are applied to them.

To achieve a reliable joint in critical applications, it is important
that the correct clamping force be applied by the fastener to the joint. This is
to say, the tension in the bolt must achieve a certain value for the joint to
be properly clamped. If the bolt tension is too low, it may loosen and cause all
clamp force to be removed with attendant damage to the structure. If it is too
high, the fastener or clamped parts could fail, also causing damage to the structure.

There are no known methods for measuring bolt tension directly without
instrumenting the fastener and/or joint. Instrumenting a joint is expensive and
time consuming and therefore seldom done in mass production. Sophisticated inferential
methods have therefore been developed to estimate the bolt tension based on known
or estimated parameters of the bolted system such as the torque applied to the
fastener by the tightening system and/or the angle of advance of the fastener.
Such methods include terminating tightening when a certain torque value is reached,
a certain angle of advance is reached as measured from a defined point, when the
yield point of the joint has been reached and others.

The types of methods used have to some extent been dependent on the
types of tools used for tightening the joint. Methods in which tightening was terminated
based on both measured torque and angle values have typically required instrumenting
the tool to acquire both types of data values. These methods have usually been
used with electrically or pneumatically driven tools, where they are practical.

In rugged or very heavy duty applications, where hydraulic torque
wrenches are typically used, it is not possible, or at best highly undesirable,
to instrument the tools. In such applications, the joint has typically been tightened
by terminating tightening in response to reaching a certain torque. This avoids
the need to instrument the tool because the torque can be determined from the pressure
applied to the wrench. The pressure is a parameter which is representative of
the torque applied to the fastener, and can be measured remotely from the wrench,
typically at the pump which supplies fluid to the wrench. The pump may include
a controller for terminating the flow of fluid to the wrench when the pressure
corresponding to the desired torque value is reached.

Another difference between hydraulic and pneumatic or electric wrenches
lies in their basic operation. Pneumatic and electric wrenches typically can rotate
the fastener during tightening for 360° or much more without stopping, until the
desired stopping point is reached. Hydraulic wrenches, on the other hand, are usually
operated by a reciprocating hydraulic piston/cylinder device operating through
a ratcheting mechanism to turn a socket for the fastener a fixed number of degrees,
e.g., 32°, each full advance of the piston. Advance of the fastener, and therefore
advance of the associated angle and torque, are in stages, with the advance starting
and stopping several times in the course of tightening a single fastener, until
the final stopping parameter, typically a final pressure, is reached.

Thus, in operation a hydraulic torque wrench socket driver will turn
for a certain number of degrees while applying torque to the fastener until it
reaches its limit of advance or until the final pressure is reached. If the stroke
reaches its limit before the final pressure is reached, the operator of the wrench
trips a switch which operates a valve to dump the wrench pressure to tank, allowing
the wrench to return to its starting point, by ratcheting around the socket. During
the resetting of the wrench, the driven socket of the wrench does not rotate but
may recede a small amount due to clearance between the socket and the head of
the threaded fastener.

Thus, as the torque wrench tightens the fastener, there is generated
a time sequence of torque pulses, each covering a limited angle (e.g., 32°), which
causes the fastener to rotate and therefore become tensioned. The space between
the torque pulses, when the dump valve is open, is used for resetting the socket
driver. The result of this complex operation is that there is a rather severe
discontinuous functional relationship between the torque, pressure or other force
dependent variables of the system with respect to the angle of advance of the fastener.
This exacerbates the problem of applying known fastener tightening methods to
the operation of a hydraulic torque wrench.

In the past, the output of hydraulic torque wrenches has been largely
controlled by monitoring and regulating the magnitude of applied hydraulic pressure.
It is well known in the art of threaded fasteners that because of variations in
the coefficients of friction at the threaded engagement and at other sliding surfaces,
the tension level (i.e., the clamping force) achieved at a given pressure (torque)
level can vary as much as 30%. More sophisticated tightening methodologies are
known, such as the "turn-of-the-nut" method disclosed in U.S. Patent No. 4,106,176,
which yield a more accurate clamping force, but require the measurement of angle
as well as torque, and have not found practical application in fastener tightening
by torque wrenches.

EP-A-0 297 515, which is considered to represent the closest prior
art, discloses a pumping unit for supplying hydraulic fluid to a hydraulically
powered torque wrench, wherein torque and angular displacement transducers are
velocated away from the moving parts of wrench to reduce the maintenance problems
of these transducers due to hansh environment.

__Summary of the Invention__
This invention provides an improvement to a pumping unit which is
adapted for supplying a flow of hydraulic fluid under pressure to a hydraulically
powered torque wrench according to the characterizing portion of claim 1. The pumping
unit of the invention further comprises a speed transducer for generating a speed
signal representative of the angular speed of the shaft and a controller arranged
for processing said speed signal into a signal which is representative of an angular
position of said wrench. The speed signal can be converted into a measure of the
pump flow rate, to avoid instrumenting the tool and provide an accurate measurement
of pump flow rate or pump flow over a given time.

This invention is useful as applied to a hydraulic torque wrench
fastener tightening system. In so doing, data representative of the torque and
angle of turn of the fastener is obtained, which can be used to monitor the tightening
of the fastener or determine a final stopping point for terminating tightening.
The invention accomplishes this without adding any attachments to the hydraulic
torque wrench.

In one aspect, pressure is measured and processed into a parameter
representative of torque and an angle parameter representative of the angle of
rotation of the fastener by the wrench is determined from a measurement of the
volume of fluid supplied to the wrench. The angle parameter is derived from pump
speed (or flow rate derived from pump speed) integrated over time. Pump speed is
measured without instrumenting or otherwise altering the wrench.

The wrench may be of the common type driven by a reciprocating piston
and cylinder device through a ratchet drive mechanism. If so, the torque and associated
angle data points define a function which in graphical form of associated pressure
and angle is defined in part by a series of spikes separated by ramps and angle
advances. Each spike begins at a first pressure which occurs just prior to the
wrench reaching a limit of advance and has a maxima and minima. Each ramp begins
at the spike minima of the previous spike and continues to a second pressure approximately
equal to the first pressure. Each corresponding angle advance, which is the set
of data points which results from turning the fastener, begins at the second pressure
and continues to the first pressure of the succeeding spike. The data points of
the spike and of the ramp are discarded, and the data points of the angle advances
are smoothed to create a characteristic function of parameters representative of
torque and angle for the joint.

The invention can be practiced with a single acting or a double acting
torque wrench, the signal processing being somewhat different depending on which
type of wrench is used. In addition, the system may be provided with a calibration
fixture to determine the volumetric rate of angle advance and the pressure vs.
torque relationship for a given wrench.

These and other objects and advantages of the invention will be apparent
from the detailed description and drawings.

__Brief Description of the Drawings__

- Fig. 1 is a plan view of a hydraulic fastener tightening system of the invention;
- Fig. 2 is a cross-sectional view of a prior art wrench of the type illustrated
in Fig. 1;
- Fig. 3 is an electro-hydraulic schematic diagram of the system of Fig. 1;
- Fig. 4 is a view similar to Fig. 3 but of an alternate embodiment;
- Fig. 5 is a graphical representation of pump flow versus pressure for a typical
hydraulic torque wrench system;
- Fig. 6 is a graph of torque versus rotation angle for a typical threaded fastener;
- Fig. 7 is a graph of pressure versus time for a hydraulic wrench tightening
system; and
- Fig. 8 is a graph of torque versus angle for a hydraulic wrench tightening
system.

__Detailed Description of the Preferred Embodiments__
Fig. 1 illustrates a system 10 of the invention which includes a
pumping unit 12, a hydraulic wrench 14 and a hydraulic line 16 connecting the unit
12 to the wrench 14 for supplying pressurized hydraulic fluid to the wrench 14
and returning the fluid from the wrench 14 to the pumping unit 12.

The wrench 14 may be of any suitable type. One such type is shown
in Fig. 2, which is of a prior art design. The wrench 14 is designed for extremely
rugged and heavy duty service, having a solid steel body 20 which houses a sleeve
22 and plug 24 which define a hydraulic cylinder 21 within the body 20. Piston
26 is slidably received in the cylinder 21 to reciprocate axially as hydraulic
fluid is introduced to the cylinder 21 at the left end of piston 26 (as viewed
in Fig. 2) and relieved therefrom via line 16.

At its rightward end, the piston 26 has a ball and socket joint in
which ball 28 is slidably received, which slidably mates with crown 30 of lever
32. Piston 26 is returned to its retracted position by compression spring 34.
A fine-toothed spline drive ratchet pawl 36 engages teeth on the outside of quill
shaft 38, which is journaled in body 20, to rotate the quill shaft 38 clockwise
as viewed in Fig. 2. On the return stroke, the ratchet pawl 36 chatters in reverse
over the teeth of shaft 38 under the bias of spring 34, in well known manner. Quill
shaft 38 drives a socket 40 (which may be removable and replaceable, as is well-known)
which engages a head of a fastener to rotate and tighten the fastener.

The unit 12 also includes a controller 18 and an automatic calibration
station 19. The unit 12 has a fixed displacement pump 13 driven by a prime mover
15 (such as an electric motor) through appropriate mechanism (not shown, e.g.,
a suitable drive mechanism such as a belt and pulley arrangement, chain and sprocket
arrangement, gear arrangement etc.) housed within the housing 17. The pump 13
may also be a two stage pump, with one stage being a low pressure variable displacement
pump (e.g., a gerotor type pump) and the second stage being a fixed displacement
pump (e.g., a piston type pump). At the higher pressures at which torque wrenches
are typically operated in the linear tensioning range of a fastener, such pumps
are fixed displacement devices.

Fig. 3 graphically depicts the system 10 in electro-hydraulic schematic
circuit diagram form. The wrench 14 is schematically illustrated as a ratchet lever
32 and single acting spring return cylinder 21, which is equivalent to the mechanism
of Fig. 2. The pumping unit 12 electro-hydraulic circuit includes the pump 13,
motor 15, a shaft 11 illustrated schematically as connecting the motor 15 to the
pump 13 and a reservoir R shown in three places, it being understood that these
are one and the same reservoir. The circuit of the unit 12 also includes a three-position,
three-way valve 45, a pressure transducer 47, a revolution counter, tachometer
or speed transducer 49, a flow rate transducer 51, relief valve 53 and controller
18, and wires 56, 58, 60, 62, 64, 66 and 68 (which may be wire pairs or any number
of wires necessary for each component) connecting the various electrical components
of the pumping unit 12 to the controller 18. Controller 18 has power cord 70 for
plugging into a wall outlet or extension cord for power to the unit 12.

The controller 18 would typically have an on/off switch 18a, and
may be provided with digital readouts 18b and 18c of pressure and pump speed, total
flow or flow rate. A remote control (not shown) may also be provided for the operator
of the wrench 14 to turn the pumping unit 12 on or off without having to walk back
to the pumping unit 12 from where he is tightening the threaded fastener. The
pressure signal, which is representative of the fluid pressure supplied to the
wrench 14 and may be displayed on digital display 18b, is processed from the signal
generated by transducer 47.

For a fixed displacement pump, each revolution of the pump drive
shaft results in a certain volume of fluid being pumped. Therefore, the pump speed,
which is measured in revolutions per minute, is representative of the flow rate
delivered by the pump. The pump speed, or the flow rate derived from pump speed,
or any other value representative of them derived from the measured pump speed,
may be integrated (or added) to yield the total flow delivered over a certain
period of time. Either the pump speed, the flow rate, the total flow or the angle
of advance may be displayed on digital display 18c, as processed from the signal
produced by transducer 49 as more fully described below.

If the pump 13 is a fixed displacement device as is preferred, the
output signal of the transducer 49 is representative of both speed and flow rate.
Furthermore, if the pump 13 is operated at a constant speed, for example by a
closed loop speed control system for the pump motor 15 or by a synchronous AC motor,
then the flow rate is constant and the total flow delivered is proportional to
time. In this case, it would be possible to determine the angle of advance of
the wrench 14 from a measurement of time, thereby making the transducers 49 and
51 unnecessary. Thus, a data acquisition system can be employed to sample the
data at a known rate. The time variable can be inferred from the number of samples
and the sampling rate, to indicate the total flow delivered to the wrench 14 for
the relevant portions of the tightening cycle when the fastener is being advanced,
as described below.

In the preferred system, in which a pump speed signal is used as
representative of flow rate, the transducer 51 is optional and is provided as a
check on the output of the transducer 49.

Since hydraulic fluid is for all practical purposes incompressible,
there is a direct relationship between the flow output of the pump 13 which is
delivered to the wrench 14 and the angle of advance of the wrench 14. Hence, the
output of the transducer 49, which is representative of pump speed and therefore
flow rate, determines the rate of advance of the wrench 14. This output can therefore
be integrated to determine the angle of advance of the fastener. As noted above,
if the pump 13 is driven at a fixed speed, so as to produce a constant rate of
advance of the wrench 14, then time (including a count representative of a clock
measurement of time) may be integrated over the periods that the fastener is actually
being advanced to yield the angle of advance of the fastener.

The relationships between speed, time, pressure and angle for a hydraulic
torque wrench are mathematically described as follows:

If F_{W} is flow to the wrench, F_{P} is flow from
the pump and F_{L} is leakage flow for the periods that the fastener is
being advanced, then
F_{W} = F_{P} - F_{L}.

The pump motor speed S is related to the pump flow F_{P}
as
follows:
F_{P} = aS,
where "a" is a constant for the specific pump and motor.

The pressure P is related to the leakage flow F_{L} as follows:
F_{L} = bP,
where "b" is a constant for the specific pump.

Combining equations (1), (2) and (3):
F_{W} = aS - bP

For hydraulic torque wrenches, the input fluid flow is proportional
to the speed of rotation of the wrench socket. That is:
F_{W} = c d&thetas;/dt
where "c" is a constant for the wrench, referred to herein as the volumetric rate
of angle advance.

If data is sampled at a high rate in comparison to the rate of change
of the variables of the system, as would be the case in the preferred embodiment,
equation (5) can be very accurately approximated by:
F_{W} = c Δ&thetas;/Δt
where Δt is the sampling period and &thetas; is the angle of the socket.

Combining equations (4) and (6) and rearranging yields:
Δ&thetas; = (aS/c - bP/c)Δt

The sample period is Δt and the torque wrench power stroke
time ts is broken up into n segments of Δt each so that ts = Δt + Δt
+ Δt + Δt + ...Δt = nΔt. At each sampling instant, data
corresponding to speed S_{i} and pressure P_{i} is taken and recorded.
Thus, for the first time interval:
&thetas;_{1} = Δ&thetas;_{1} = (aS_{1}/c
- bP_{1}/c)Δt

In general for any time interval Δt:
&thetas;_{i} = Δ&thetas;_{i} = (aS_{i}/c
- bP_{i}/c)Δt

Finally, the total wrench angle &thetas; at time t_{1}, time
t_{2}
and at any time t_{n} can be found as follows:
*&thetas;* (t_{1}) = &thetas;_{1}
&thetas; (t_{2}) = &thetas;_{1} + &thetas;_{2}
..., so that
&thetas; (t_{n}) = &thetas;_{1} + &thetas;_{2}
+ ... &thetas;_{i} + ... &thetas;_{n}

Thus, knowing the time variable, the speed variable and the pressure
variable provides the angle variable of the torque wrench. As stated above, if
the speed is constant, then only the time and pressure variables need to be known
to yield angle. Although measuring the flow rate directly dispenses with both of
the time and speed variables, it is more problematic to measure. Also, if leakage
is relatively small, it can be neglected, so pressure need not be known to yield
an accurate determination of angle.

As shown in Fig. 3, in the at-rest position of the solenoid valve
45, flow from the pump 13 is directed to the reservoir and backflow from the wrench
14 is blocked. When solenoid 45a is actuated by controller 18, the valve 45 is
shifted rightwardly to communicate the entire output of pump 13 to the cylinder
21 of wrench 14, thereby causing piston 26 to advance, or if it has reached its
limit of advance (i.e., as far as it will go), causing the pressure in the cylinder
21 to increase sharply, the rate of increase depending on the volumetric stiffness
of the hydraulic system, which is typically very stiff.

Since the system is very stiff, when the pressure limit of the relief
valve 53 is reached, which is set to be higher than any pressure that might be
attained in normal tightening of the fastener during a stroke of the wrench 14,
the valve 53 opens to relieve the pressure in cylinder 21 to the reservoir (essentially
zero pressure). In this position, output from the pump 13 is also directed to the
reservoir. The spring 34 thereby returns the lever 32 to its starting, fully retracted
position.

Alternatively, if the relief valve 53 was not provided, the solenoid
45a could be de-energized and solenoid 45b energized by controller 18, so as to
shift the valve 45 leftwardly as viewed in Fig. 3, to relieve the pressure in
cylinder 21 to the reservoir and allow the lever 32 to return under the influence
of the spring 34.

Controller 18 is programmed to only collect pressure and flow rate
data, as measures of torque and rate of angle of advance respectively, during the
periods that the fastener is actually advancing in angle. Fig. 6 is an idealized
graphical representation of the torque versus angle function for the tightening
of a typical fastener. An idealized graphical representation of pressure versus
time is shown in Fig. 7 for the tightening system of Figs. 1 and 3, utilizing
a ratcheting type hydraulic torque wrench of the type illustrated in Fig. 2. Fig.
8 illustrates torque (the product of pressure and a constant conversion factor)
versus actual measured angle for tightening a fastener with a ratchet type hydraulic
torque wrench. Points on the graph of Fig. 8 corresponding to points on the graph
of Fig. 7 are identified with the same letters.

The torque-angle curve of Fig. 6 may be viewed in four segments.
Segment 80 is a range of initial tightening in which the parts of the joint are
brought together without significant clamping and is generally linear and of a
low slope. The next portion 82 is the snug or clamp-up range in which the mating
threads of the fastener become seated and initially stressed, and the torque angle
gradient changes from its previous low value to a significantly higher value which
stays substantially constant over the bolt tensioning range 86. Compression of
gaskets or other parts of the joint having a significantly lower stiffness than
the fastener occurs by the end of portion 82. Beyond the linear bolt tensioning
range 86, the non-elastic yield region 88 occurs, in which the fastener or clamped
parts of the joint yield plastically. Point "V" represents the desired stopping
point for tightening the fastener, which is on the linear part of the torque angle
curve, below the yield point of the joint.

The pressure-time curve of Fig. 7 differs dramatically from the torque-angle
curve of Fig. 6. However, it is possible to process the pressure-time curve of
Fig. 7 to approximate the torque-angle curve of Fig. 6.

To process the pressure versus time data so that the discontinuities
are removed and a smooth torque-angle curve is obtained, starting at the beginning
of the first stroke, at point A, the pressure and speed data is recorded until
the end of the first stroke, at point B. The pressure signal and speed signal
are in the form of electrical output signals from the respective pressure 47 and
speed 49 transducers, which may be converted (if necessary) by a suitable analog
to digital converter in the controller 18 into corresponding digital signals. These
signals are converted by the controller into respective torque and angle values,
for example, by comparing the digital output values in a look-up chart to determine
the corresponding torque and angle values, which can be used to establish a point
on the graph of Fig. 8. The flow rate value is first integrated to yield the total
flow since the onset of advance, or to yield the incremental flow to the wrench
which is added to the previous flow to the wrench, before looking up the corresponding
incremental angle value in the look-up chart. The incremental angle value is the
angle traversed since the beginning of the present stroke of the wrench 14, which
can be added to the angle traversed on the previous strokes to yield the total
angle of advance.

Alternatively, the output signals may be mathematically processed
to yield corresponding torque and angle values. The conversion of pressure to torque
is relatively straightforward mathematically, if the moment arm of the piston
26 acting on the socket 40 is constant, as it may be assumed to be with reasonable
accuracy for many hydraulic wrenches. In that case, pressure can be converted
to torque by multiplying it by a suitable conversion factor, which is constant,
and suitable adjustments made to the value to account for friction (if applicable)
and the force due to the compression of spring 34. For example, if spring 34 has
a significant spring rate, then part of the pressure force must be attributed to
compressing the spring 34 and that part increases as the piston 26 advances and
the spring 34 becomes compressed. In that case, the conversion of pressure to torque
desirably takes into account the spring force, which varies according to the compression
of the spring 34, i.e., according to the incremental angle of advance of the fastener.
As stated above, angle may be determined from the speed, time and pressure measurements,
using equation (9).

With either the look-up table or the calculation method, calculation
times are not significant in comparison with the tightening process time, since
tightening with the hydraulic wrench system is a start and stop process with periods
in which the fastener is not being turned when the wrench is being reset, which
periods provide ample calculation time. The raw data thus obtained (or obtained
by using the look-up table approach) may be processed by any desired means to
yield a smooth curve or function, for example by a least squares fit smoothing
technique.

Referring to Figs. 7 and 8, angle advance segment A-B of the first
stroke, and corresponding segments F-G, K-L, P-Q, and U-V of the subsequent respective
second, third, fourth and fifth strokes, represent actual turning of the fastener
by the wrench 14. Point B, and corresponding points G, L and Q of subsequent cycles,
represent the point in the stroke cycle of the wrench 14 in which the piston 26
is fully extended and bottomed in the cylinder 21, i.e., at this point the wrench
14 is at its limit of advance. Advance of the fastener stops at that point and
the result of continuing to pump fluid to the wrench 14 is only to increase the
pressure in the cylinder 21 at a high rate.

As stated above, the pressure relief valve 53 opens at a certain
pressure limit P_{L}, shown in Fig. 7, which is above any possible normal
pressure at the point at which tightening is terminated. When a pressure equal
to or greater than the pressure limit P_{L} is detected, the valve 53
dumps pressure from the cylinder 21 and from the pump 13 to the reservoir, thereby
allowing the wrench 14 to reset under the bias of spring 34. In Fig. 7, the pressure
limit P_{L} is reached at point C for the first stroke and at points H,
M, and R for the respective second, third, and fourth strokes.

The part of the curve in Figs. 7 and 8 from points C to D represents
the resetting of wrench 14, as does the portions H-I, M-N, and R-S for the respective
second, third, and fourth strokes. At points D, I, N and S, the piston 26 has
retracted to its fully retracted position, i.e., to its limit of retraction, in
which lever 32 is at its zero degree incremental angle starting point. Point D
for the first stroke, and points I, N, and S for the respective second, third,
and fourth strokes, represent essentially zero pressure, i.e. full resetting of
the wrench 14 back to the zero degree incremental angle starting point. This triggers
the valve 53 to close, thereby repressurizing the wrench 14. Referring specifically
to Fig. 7, the segment from D-E, and the corresponding segments I-J, N-O and S-T,
are due to time delay needed to process the data and begin the next stroke.

Ramp segment D-F for the first stroke, and ramp segments I-K, N-P,
and S-U, for the respective second, third, and fourth strokes, represent the build-up
of pressure in the cylinder 21 without advancing the fastener angle. In going
from points B to C to D and then from D to F, a change in angle is illustrated
in Fig. 8, negative going from B to C to D and positive going from D to F. However,
this is small (e.g., 4-5°) and only accounts for clearances within the mechanism
of the wrench 14 and between the socket and fastener head. The fastener itself
does not rotate backwardly or advance significantly during this portion of the
cycle.

The data points defining the spike B-C-D and defining the segment
D-F are discarded, since they are meaningless to the rotation of the fastener and
only represent resetting of the wrench 14. The same is true for the segment G-K,
L-P and Q-U for the respective second, third, and fourth strokes of the wrench.

The slope of the segment B-C, and the corresponding segments G-H,
L-M, and Q-R for the second, third, and fourth strokes, respectively, is nearly
infinity, and therefore is distinguishable from any normal slope of the torque-angle
curve. Therefore, the points B, G, L, and Q may be determined during tightening
by sensing the onset of this very high slope. For example, a running average calculation
of the slope obtained from the data points may be compared to a certain slope maximum,
which value is chosen to be above the highest expected slope of the bolt tensioning
range of the torque angle curve. When the running average slope becomes greater
than the slope maximum, the data begins to be discarded. Alternatively, since
point C occurs at essentially the same time as point B due to the incompressibility
of hydraulic fluid, the data may begin to be discarded when the pressure limit
P_{L} is detected, or counting back a certain number of data points before
then.

From the point B, and the corresponding points G, L and Q of the
respective second, third and fourth cycles, the data may continue to be discarded
until the pressure at these points is once again obtained, less a correction factor.
Thus, point F, where data acquisition restarts, and the corresponding points K,
P and U, may be somewhat below their respective corresponding points B, G, L and
Q. Part of the difference between the points B and F, between the points G and
K, between the points L and P, and between the points Q and U is due to the fact
that at the previous point B, G, L, or Q, the spring 34 is fully compressed (since
the wrench is at its limit of advance) and at points F, K, P and U the spring is
at its least compression (since the wrench is at its limit of retraction). Part
of this difference is also due to the socket tightening against the head of the
fastener prior to the fastener actually starting to turn. Thus, one may either
correct for the difference between the points B and F, and the corresponding other
points, by adding an appropriate factor to the point B accounting for the lack
of spring compression and the prestressing of the fastener prior to turning, or
may use another smoothing technique in this part of the curve, to fit the data
points to the relatively flat and straight curve which is expected in this part
of the curve. Alternatively, in some applications it may be acceptable to simply
restart data acquisition when the pressure is equal to the pressure at which data
acquisition last terminated, and join the curve segments with a straight line
or use another smoothing technique.

This procedure is applied for each of the strokes of the wrench 14
until the final stopping parameter is obtained, to stop at point V. In the curves
shown in Figs. 7 and 8, this occurs during the fifth stroke prior to reaching
the pressure limit P_{L}. The parameters which define the stopping point
V may be determined by any desired tightening methodology, preferably one that
relies upon values dependent upon both torque and angle, to fully realize the
benefits of the invention. The final stopping parameter is obtained by manipulating
the data points collected as described above, and when that stopping parameter
is obtained, at point V (or slightly before), the controller 18 sends a signal
to de-energize solenoid 45a, which returns valve 45 to its center position, thereby
terminating tightening so that the fastener stops at point V.

One such tightening methodology is described in U.S. patent 4,106,176.
This is a modified turn-of-the-nut methodology in which a fixed angle, empirically
determined for the particular joint being fastened, is measured from the zero
torque intercept &thetas;_{0} (Fig. 6) of the bolt tensioning portion of
the torque angle curve. In practicing this methodology in connection with the present
invention, torque and angle values for the joint being tightened are determined
from the measured pressure and speed data obtained, the bolt tensioning range of
the torque angle characteristic curve is extrapolated down to the zero torque
axis, and the final stopping angle &thetas;_{v} (Fig. 6) (which may be
easily converted to a time or flow value) or torque (which may be easily converted
to a pressure value) is added to the corresponding value at the zero torque intercept
to determine the final stopping parameter, which may be expressed in terms of torque,
pressure, angle, time, flow or rotations of the pump shaft, for the period(s)
during a stroke of the wrench. The instruction to terminate tightening is then
issued by the controller 18 to stop tightening when the final stopping parameter
value is reached.

Other methodologies may also be used to practice the invention, such
as the yield point method, in which the yield point of the joint is determined
based on the measured values indicative of torque and angle and tightening is
terminated in response thereto, or turn of the nut as measured from a certain pressure
or torque. Other methods utilizing torque and angle values may also be applied
in practicing the invention, or the invention may simply be applied to monitor
torque and angle parameters during the tightening process, with the operator terminating
tightening if they deviate from the expected in the operator's judgement.

There is some leakage in the flow from the pump 12 to the wrench
14, which increases with pressure. Therefore, not all the flow delivered by the
pump 12 actually rotates the fastener, a small amount of it being sacrificed to
leakage. Leakage increases approximately linearly with pressure, as illustrated
in Fig. 5, so a suitable correction factor can be employed if the angle of the
fastener is mathematically determined from the pressure and flow rate data (See
equation (9)). Alternatively, the angle of advance of the fastener can be determined
in a look-up chart relating, for example, pressure and total flow, pressure and
the total number of revolutions of the pump 13 or pressure and time, with flow,
revolutions or time measured from the start of each stroke of the wrench 14.

An alternate hydraulic schematic for the pumping unit 10 is illustrated
in Fig. 4. The circuit of Fig. 4 is substantially identical to that in Fig. 3 and
corresponding elements are identified with the same reference number, plus a prime
(') sign. The only difference between the wrench 14' and the wrench 14 is that
the wrench 14' is not a single acting spring return wrench, but is a double-acting
wrench, which is returned by hydraulic pressure, as illustrated in cylinder 21'.
Accordingly, the solenoid valve 45' in Fig. 4 is a four-way, rather than three-way,
valve, since hydraulic pressure is used to return the wrench to its limit of retraction
after each stroke. Thereby, the effects of compressing the spring 34, and the
effects which it has on the pressure, are avoided in the embodiment of Fig. 4.

Summarizing with reference to Fig. 7, a signal processing algorithm
for practicing the invention is as follows:

- 1. Starting at A, sample and record the data until the end of stroke B. The
end of stroke may be detected by monitoring the pressure limit signal P
_{L},
since point C is virtually at the same time as point B. This power stroke covers
the time interval from t0 to t1. Multiply the P variable by a correction factor
to convert from pressure P to torque T. For a single acting wrench, also subtract
out a value attributed to the return spring. No return spring correction is needed
for the double acting wrench. This segment is now part of the torque versus time
curve. Using equation (9) above, convert the time axis variable (t) into an angle
variable (&thetas;) axis.
- 2. Data from B to D is ignored as this is part of the resetting of the wrench.
That is, data from time t1 through t2 is to be discarded.
- 3. Data from D to E is ignored as this is due to the delay needed to process
data and begin the next stroke. That is, data from time t2 through t3 is ignored.
- 4. At F, the pump begins its next stroke. Data taken from E to F is ignored
as this data is due to pump pressure build-up to the prior pressure level. If the
points B and F do not quite match in pressure, then average or interpolate the
curve at this point to make it smooth.
- 5. Data from F to G is the next power stroke segment. This is time segment
t3 through t4. Treat this segment as in Step 1 above. After the conversion to T
versus &thetas; as described in that step, append it to the previous T versus
&thetas; segment.
- 6. Repeat Steps 2 through 5 until the desired stopping point is reached, using
any suitable tightening methodology.

Thus, the data from t1-t3, t4-t6, t7-t8 and t10-t12 is discarded
and the remaining data from t0-t1, t3-t4, t6-t7, t9-t10 and t12-t13 is put together
and converted to torque and angle values to yield a curve which approximates the
curve of Fig. 6, up to the stopping point V.

The invention may be practiced with any suitable hydraulic wrench,
but it is important to know the characteristics of the particular wrench being
used. To this end, an automatic calibration fixture 19 may be provided as part
of a pumping unit 12. The wrench 14 being used is hydraulically connected to the
pumping unit 12 and then placed on the automatic calibration fixture 19, which
has a rotary head 19a with which the socket of the wrench 14 is engaged. The head
19a is rotated by operating wrench 14, and a rotation sensor 19b of the unit 19
measures the rotation of the head 19a by the wrench 14. A torque sensor (not shown)
may also be employed in the unit 19 to measure the torque exerted on the head 19a
by the wrench 14. If so, the head 19a may be rotated with increasing resistance
up to the pressure limit P_{L}, and the measured values of pressure, pump
speed, angle of advance and torque can be related in two look-up tables, one relating
pressure and angle to torque, and the other relating the integral of pump speed,
i.e., revolutions, (or a value representative thereof such as the integral of flow
rate, i.e., total flow delivered to the wrench, or time if constant speed) and
pressure to angle of advance. Thereby, look-up tables for the torque and angle
produced by the wrench 14 as a function of the parameters measured by the pumping
unit 12 in operation (i.e., pressure and flow rate or rpm or time) can be automatically
generated by the pumping unit 12 for the particular wrench 14.

Alternatively, if the calculation method is used to convert pressure
to torque and time to angle, the angle values measured by the fixture 19 and the
flow delivered to the wrench 14 to produce the measured advance angle (as determined,
for example, from the output of sensor 49 and a measurement of time, See equation
(9)) can be used to determine the angle of rotation per unit volume of flow to
the wrench (i.e., the volumetric rate of angle advance, c in equation (9)) for
the particular wrench being used.

If torque is also measured by the unit 19, the slope of the torque
vs. pressure relationship can be determined and applied subsequently to determine
torque from the pressure measurements when tightening fasteners. The leakage correction
is more a characteristic of the pump and so can be assumed to be constant from
wrench to wrench. If a single acting wrench is used, the pressure due to the reaction
force of the return spring can also be determined, for example, by shifting valve
45 to its center position at or near the fully extended position of the wrench
(with no torque exerted on the socket 19a) and measuring the pressure exerted
by the spring 34.

Depending upon the operating pressure, some amount of pump flow which
does not directly rotate the wrench may be attributable to the elasticity of the
hoses and other components and the compressibility of the fluid. If this is significant
in the application to which a system of the invention is applied, this should be
accounted for and an appropriate correction made. If a look-up table is used to
determine the angle values, then no correction would be needed because the correct
angle associated with a certain pressure and time, flow or number of revolutions
of the pump would be built into the table. Such a table could be automatically
generated using the calibration fixture 19. If a calculation method is employed,
correction factors can be determined using fixture 19 by running it through two
cycles: one being a non-movement cycle where the system measures oil volume due
to system component expansion, fluid compressibility and leakage (at one or more
operating pressures); and a second cycle, which could be done at low pressure,
in which the volume of oil used to extend the wrench for one full cycle is determined.
These values can then be used to correct the calculated values for system expansion,
fluid compressibility and leakage characteristics.

It is also noted with respect to Figs. 7 and 8 that in practicing
a certain tightening methodology, the portion of the pressure angle curve leading
up to point U, and slightly beyond point U, may be irrelevant. If so, all data
prior to that point may be discarded, and only data subsequent to that point, determined
by setting a certain minimum pressure combined with a slope within the expected
range of slopes of the bolt tensioning portion of the torque-angle or pressure-angle
curve, need be determined. For example, in the modified turn-of-the-nut methodology
referred to above in U.S. patent 4,106,176, only the linear bolt tensioning range
of the curve is of interest, which could be deemed to start at a certain pressure
level which is chosen to be above the lowest expected pressure of the bolt tensioning
range but below the expected final stopping point.

Many modifications and variations to the preferred embodiment as
described will be apparent to those skilled in the art. For example, a system of
the invention could be programmed to retract by operating valve 45 or 45' at a
certain angle of rotation from the beginning of each stroke so as not to fully
extend the wrench piston, which would avoid the pressure spikes and result in quieter
operation of the wrench. Also, many diagnostics could be programmed into the system,
for example, a warning could be generated if the pressure limit was detected before
enough flow had been delivered from the beginning of a stroke to produce a full
stroke of the wrench, which would indicate that either the wrench had not fully
retracted after the last stroke or that abnormal resistance was being encountered
in tightening.