Downloaded from http://onepetro.org/PO/article-pdf/10/01/7/2602443/spe-26525-pa.pdf/1 by Robert Durham on 20 September 2023
`
`Field Application of Hydraulic Impedance
`Testing for Fracture Measurement
`
`R.W. Paige. LIt Murray. and J.D.M. Roberts. SPE. BP Exploration
`
`Summary
`Hydraulic impedance testing (HIT) is a technique for detecting and
`measuring formation fractures intersecting wellbores. A pressure
`pulse is introduced into a well, and the resulting pressure trace is in(cid:173)
`terpreted to give fracture dimensions. The first part of this paper de(cid:173)
`scribes how HIT can be used to estimate fracture dimensions and
`presents some results from a laboratory experiment that show that
`dimensions can be measured accurately with HIT. The remainder of
`the paper describes field examples of the application of HIT. A dem(cid:173)
`onstration of how HIT traces change as pressure is reduced, which
`provides a method for determining fracture closure pressure, is in(cid:173)
`cluded.
`
`Introduction
`The presence of fractures that enhance well productivity or injectiv(cid:173)
`ity can dramatically improve oilfield profitability. It is therefore im(cid:173)
`portant to understand how fractures behave so that fracture designs
`and production strategies can be optimized.
`HIT is a technique for detecting and measuring the size of frac(cid:173)
`tures that communicate with wellbores. It can therefore be an impor(cid:173)
`tant tool in the drive to improve our understanding of fracturing and
`to monitor fracture growth.
`HIT uses the transient response of the fluid in the wellbore and
`fracture that results from the introduction of a pressure pulse into the
`well to provide information about the fracture. The principles be(cid:173)
`hind the technique are not new. In the 1960's, Anderson and Stahl!
`reported changes in the period of fluid oscillation in a wellbore as
`a fracture formed. In the 1980's, Holzhausen published several pa(cid:173)
`pers2-4 detailing a form of HIT, although the method for analyzing
`pressure traces differs from that used for the work covered here.
`The method reported here provides estimates of both fracture
`height and fracture length for open fractures that do not contain
`proppant. This can provide a useful addition to the tools available
`for fracture measurement, particularly in the design of hydraulic
`fractures where the engineer often has a good idea of fracture face
`area and a relatively poor estimate of fracture height.
`This paper starts by describing the method developed for estimat(cid:173)
`ing fracture dimensions. Results of a laboratory study that show that
`HIT accurately measures known fracture geometries follow. Field
`investigations of fracture opening and closing with hydraulic im(cid:173)
`pedance testing are then presented.
`
`HIT Method
`Fig. 1 is a schematic of the essential features of HIT. A pressure
`pulse is introduced into the top ofthe well. The pulse travels down
`the well and is reflected, for example at the mouth and tip of any
`fracture communicating with the well. The reflected pulses travel
`back to the surface, where they are detected by a pressure transduc(cid:173)
`er. The pressure trace thus obtained can then be interpreted to esti(cid:173)
`mate the size of any fracture connected to the well.
`The equipment used to take the measurements consists of a device
`for introducing a pressure pulse into the well and a high-frequency
`pressure transducer connected to suitable recording equipment. In
`its simplest form, the pulse generator may be a ball valve connected
`to the wellhead and exhausting to atmosphere, which can be rapidly
`
`Copyright 1995 Society of Petroleum Engineers
`
`Original SPE manuscript received for review Oct. 3, 1993. Revised manuscript received
`Sept. 6, 1994. Paper accepted for publication Oct. 10, 1994. Paper (SPE 26525) first pres(cid:173)
`ented at the 1993 SPE Annual Technical Conference and Exhibition held in Houston, Oct.
`~.
`
`opened and closed manually. Alternatively, mechanically con(cid:173)
`trolled devices that generate shorter, reproducible pulses may be
`used. The pressure transducer may be attached to the wellhead,
`avoiding the use of any downhole equipment.
`
`Interpreting HIT Traces. Fig. 2 shows the form of the pressure
`trace generated by HIT. The trace shows an initial pulse (A), a re(cid:173)
`flection from the fracture mouth (B), and a reflection from the frac(cid:173)
`ture tip (C). The observation that the fracture response comprises re(cid:173)
`flections from the fracture mouth and the fracture tip allows a
`method to be developed for estimating fracture dimensions.
`Fracture height can be estimated from the magnitude ofthe reflec(cid:173)
`tion at the fracture mouth; fracture length can be determined from
`the time that the pulse takes to traverse the fracture. Use of the dis(cid:173)
`tinct reflections from the fracture mouth and fracture tip distin(cid:173)
`guishes the method developed by BP from that adopted by Holzhau(cid:173)
`sen.2-4
`When a pulse travelling down a pipe encounters a change in hy(cid:173)
`draulic impedance, it normally produces a reflected pulse and a
`transmitted pulse. When the hydraulic impedance increases, both
`the reflected pulse and the transmitted pulse are in the same sense
`as the incident pulse. When the hydraulic impedance decreases the
`reflected pulse is inverted. Thus, in Fig. 2, Pulse B was generated
`when Pulse A encountered a reduction in hydraulic impedance at the
`fracture mouth. Pulse C was generated by a reflection at the end of
`the fracture (very high impedance) and is therefore in the same sense
`as the initial pulse. For flow in pipelines, the impedance change is
`usuall1y the result of changes in pipe diameter. The other factor that
`changes hydraulic impedance is the wave speed in the conduit.
`When the pulse hits a fracture mouth, changes in flow area and wave
`speed are important; the wave speed in the fracture is typically 100
`m1s compared with 1500 m1s in the wellbore.2
`
`Pulse Transmission in the Wellbore. Pulse transmission in the
`wellbore may be computed with standard techniques for transient
`flow in pipes. The capacitance, Cw, inertance, lw, and resistance, Rw,
`of the wellbore are given in Chap. 12 of Ref. 5 as
`
`I
`_
`lw - - - 2 '
`;rrgrw
`
`;rrgr~
`Cw = - 2 ,
`a w
`
`and Rw = ::;;",
`
`(1)
`
`(2)
`
`(3)
`
`where the resistance term given is for laminar flow (an alternative
`expression is used for turbulent flow). The resistance term is small
`relative to the inertia term at the frequencies of interest, and if resis(cid:173)
`tance is ignored, then
`
`(4)
`
`Zw = j(Iw/Cw)
`Note that while resistance has only a small effect on hydraulic im(cid:173)
`pedance, resistance effects are significant for pulse propagation
`down wellbores. Resistance effects are discussed in the section on
`Fracture Height Determination.
`The fracture can also be added to the hydraulics system by
`approximating the fracture as a ID hydraulic unit. The fracture is
`then treated as a pipe along which pulses will propagate.
`
`SPE Production & Facilities, February 1995
`
`7
`
`IWS EXHIBIT 1058
`
`EX_1058_001
`
`

`

`Downloaded from http://onepetro.org/PO/article-pdf/10/01/7/2602443/spe-26525-pa.pdf/1 by Robert Durham on 20 September 2023
`
`IWS EXHIBIT 1058
`
`EX_1058_002
`
`

`

`Downloaded from http://onepetro.org/PO/article-pdf/10/01/7/2602443/spe-26525-pa.pdf/1 by Robert Durham on 20 September 2023
`
`Pulse
`generator
`
`Pressure (psi) (Variable baseline)
`7
`
`6
`
`4
`
`A
`
`Pressure
`transducer
`
`21-.:----"""\
`
`Composite block fracture system 8" thick
`Transducers fixed in fracture face
`Blocks restrained in 30" dia. steel flange
`
`Variable profile fracture
`
`Well/fracture
`................... communication strip
`"-
`
`\
`
`\
`
`_--------
`
`...... ./' ",..
`
`/
`
`I
`
`\
`
`/
`
`II
`
`0.02
`
`0.04
`
`Fracture Tip
`0.08
`0.06
`Time (sec)
`
`0.1
`
`0.12
`
`Fig. 4-Equipment used in HIT experiment.
`
`Fig. 5-HIT experiment pressure traces.
`
`Fracture Height Determination. For a simple constant-diameter
`wellbore connected to a fracture, the reflection coefficient at the
`fracture mouth, R, is
`
`system being investigated was characterized much better than it is
`in the field.
`
`Zj - Z'"
`R = Zj + Z",'
`
`(12)
`
`where q and z,., are the hydraulic impedances of the fracture and
`wellbore, respectively. By obtaining the reflection coefficient from
`the measured field pressure trace and using known wellbore param(cid:173)
`eters, we can determine q. Substituting Eqs. 6 and 7 into Eq. 8, re(cid:173)
`arranging, and entering wellbore and rock parameters allows frac(cid:173)
`ture height to be determined from
`
`hJ = 4;,p [gZj~a~ v)r,
`
`.(13)
`
`where Pe can be determined from either HIT, as described below in
`the section on Determining Fracture Closure, or a step-rate test.
`While the effect of resistance is likely to be small at the junction
`between the wellbore and the fracture (if the resistance term for the
`fracture were large, the large inverted reflections from the fracture
`mouths would not occur), it will be significant for pulse transmis(cid:173)
`sion along the wellbore. Fig. 3 shows the decay of a pulse in a
`lIOO-m-deep, 14-cm-diameter unperforated well.
`Pulse height is reduced by approximately 10% over the 2200 m
`it travels. This effect needs to be included in estimations of the re(cid:173)
`flection coefficient at the wellbore fracture junction from pressure
`traces measured at the wellhead. In its simplest form, this can be ef(cid:173)
`fected by adjusting the value of the reflection coefficient by use of
`an attenuation factor appropriate to the pipe size and the flow condi(cid:173)
`tions.
`Using average values of capacitance and inertance for the frac(cid:173)
`ture, rather than the localized values at the fracture mouth, would in(cid:173)
`crease the calculated value of fracture height by approximately 22%
`without changing the fracture length estimate. In practice, the pulse
`length determines the region of investigation and hence the reflec(cid:173)
`tion coefficient for the fracture mouth, so the height is likely to fall
`between these two values. However, the pulse length in the fracture
`is normally less than the fracture length, so the average height of the
`zone investigated is likely to be closer to that at the wellbore than
`to the average. While this simplification affects the results, it is like(cid:173)
`ly that wellbore fracture communication has an even bigger impact
`on fracture height measurement, as discussed in the next section, so
`further refinement of this part of the model may not improve the re(cid:173)
`sults significantly.
`There are important uncertainties in the systems being modeled,
`and the analysis method described has some noteworthy simplify(cid:173)
`ing assumptions. However, it is questionable whether the system be(cid:173)
`ing investigated is defined well enough to warrant significant refine(cid:173)
`ment of the analysis method. The next section looks at a laboratory
`experiment performed to test the validity of the method when the
`
`SPE Production & Facilities, February 1995
`
`Laboratory Testing of HIT
`A laboratory study was done to ascertain how well fracture size can
`be determined with HIT. Fig. 48 is a schematic of the rig. A 40-m(cid:173)
`long wellbore of 39-mm diameter is connected to a variable-size
`fracture. The fracture is contained in the central layer of a three-lay(cid:173)
`er sandwich between two thick perspex blocks. It was necessary to
`make the fracture walls from a low-Young's-modulus material such
`as perspex so that the wave speeds in the fracture would be similar
`to those in the field. High-sensitivity pressure transducers were
`mounted in the wellbore and fracture walls to monitor pulse propa(cid:173)
`gation.
`Fig. 5 shows typical pressure traces obtained during the experi(cid:173)
`ment. Pulse A is the initial pulse; Pulse B is the pulse reflected from
`the fracture mouth; and Pulse C is the pulse from the fracture tip. The
`pulse reflected from the fracture tip can clearly be seen in the pres(cid:173)
`sure trace obtained from the fracture mouth as it enters (Point a) and
`leaves (Point c) the fracture. The pulse can also be seen at the frac(cid:173)
`ture tip (Point b), where it is magnified relative to the pulses at the
`fracture mouth because the incoming and outgoing signals superim(cid:173)
`pose.
`Table 1 compares the fracture sizes measured with the HIT meth(cid:173)
`od described above with the actual geometries for five different
`fracture sizes.
`
`Wellbore Fracture Communication. A number ofruns were made
`with reduced communication between the wellbore and fracture to
`try and establish the likely response of a deviated well. For the full
`fracture height of the first geometry given in Table 1, there were 24
`perforations. Runs were made with 8 perforations covering the cen(cid:173)
`tral third of the fracture height and with 1 perforation at the fracture
`centre. The run with communication over the central third of the
`fracture gave height and lengths of 489 and 444 mm respectively
`whereas the run with one perforation gave 168mm and 480mm.
`Taking these results with those for geometry 4 of Table 1 indicates
`
`Shape
`
`Width
`(mm)
`
`~ 1
`1
`J
`
`::::>
`1
`C)
`1
`~ 2
`
`TABLE 1
`Height (mm)
`HIT
`Actual
`483
`500
`
`Length (mm)
`HIT
`Actual
`521
`518
`
`500
`
`475
`
`166
`
`170
`360
`
`500
`
`157
`348
`
`506
`
`521
`521
`
`521
`
`177
`
`494
`479
`
`543
`
`9
`
`IWS EXHIBIT 1058
`
`EX_1058_003
`
`

`

`Downloaded from http://onepetro.org/PO/article-pdf/10/01/7/2602443/spe-26525-pa.pdf/1 by Robert Durham on 20 September 2023
`
`Pressure (psi)
`1.800
`
`1,750
`
`1.700
`
`1.650
`
`1.600
`
`1.550 OL--~2---~4-~-6~--.l.8---1LO--....I12---1-4---'
`
`lime (sec)
`
`Fig. 6-Magnus C7 HIT trace.
`
`that poor communication is likely to result in under-estimation of
`fracture height, although the measurement of fracture length is not
`likely to be affected.
`
`Field Application of HIT
`HIT has been used successfully to detect fractures in more than 50
`water injection wells. Refs. 9 and 10 report the presence of small
`fractures in the poorly consolidllted rock in the Forties field whilst
`Reference 8 indicates that larger fractures are present in more con(cid:173)
`solidated formations. HIT has been performed on several water in(cid:173)
`jection wells in the Magnus field in the North Sea. Fig. 6 shows one
`of the HIT pressure traces obtained during a pressure fall-off on well
`C7, which is near to vertical and has a constant diameter T' comple(cid:173)
`tion. This field case provides a relatively simple example of how the
`above procedure might be applied.
`
`Determining Fracture Closure Pressure. The fracture closure
`pressure may be obtained from a step rate test (Fig. 7) or from the
`HIT traces.
`If a series of HIT's are made during a pressure fall-off, the
`changes in the pressure trace will indicate changes in fracture geom(cid:173)
`etry. This provides a method for determining fracture closure pres(cid:173)
`sure and also the excess pressure, required for evaluating fracture
`dimensions. Fig. 8 gives reflection coefficients at the fracture mouth
`for a series of HITs made during a pressure fall-off on Magnus C7.
`It can be seen that the reflection coefficient increases as pressure is
`reduced, until the wellhead pressure reaches about 450psi. This is
`approximately the same pressure as that at which the gradient
`changes on the step rate test and is taken to be the fracture closure
`
`Pressure (psi)
`
`Pressure is bottom hole
`1,500 minus hydrostatic head
`
`1,000
`
`500
`
`oL-_~-_L--_-----~--------'
`
`o
`
`10
`
`20
`Flow Rate (mbd)
`
`30
`
`40
`
`Fig. 7-Step rate test on Magnus C7.
`
`pressure. After a small transition, the HIT traces measured below
`this pressure do not appear to change further, although the form of
`the trace indicates that a channel is still open into the formation.
`The gradient of the step rate plot (Fig. 7) beneath fracture closure
`pressure also supports the view that channels into the formation re(cid:173)
`main open. Channels which remain open beneath fracture closure
`pressure have been detected by HIT in many mature water injectors.
`This contrasts with HITs measured on new injectors, where frac(cid:173)
`tures close when wells are shut-in, as illustrated in Fig. 10.
`
`Estimating Fracture Dimensions. The fracture length is obtained
`from the delay time (tf= 1.8sec in Fig. 6) and the wavespeed (63 mls
`from Eq. 10). In computing excess pressure a factor of 0.5 has been
`used to take account of poroelastic effects. This gives a fracture
`length of 56m. This length is within the range of 30-1 OOm predicted
`by simulation of pressure fall-off tests (performed on Magnus well
`C2) using a computer program which couples fracture mechanics
`with a reservoir simulator!! and by conventional pressure fall-off
`analysis. Field observation also supports the view that the fractures
`in Magnus are of moderate length, as high injectivity has been
`achieved without early water breakthrough to producers, despite the
`likelihood that fractures are orientated in the direction from injector
`to producer.
`The reflection coefficient for the fracture mouth is obtained from
`the field trace (Fig. 6) as - 0.71. HIT measurements were made on
`Magnus C5 (a well with a similar completion) prior to perforation
`
`Pressure (psi)
`2,000
`
`0
`
`0
`
`0
`
`0
`
`1,500
`
`1,000
`
`500
`
`0
`
`0
`
`0
`
`0
`
`0
`
`8
`
`Depth (m)
`3,260
`
`3,280
`
`3,300
`
`3,320
`
`3,340
`
`3,360
`
`3,380
`
`3,400
`
`0
`
`-1
`
`-0.95
`
`-0.9
`-0.85
`-0.8
`Reflection Coefficient
`
`-0.75
`
`-0.7
`
`3,420
`
`0
`
`10
`
`20
`Flow Rate (mbd)
`
`30
`
`40
`
`Fig. 8-Fracture mouth reflection coefficient for Magnus C7.
`
`Fig. 9-Spinner trace for Magnus C7.
`
`10
`
`SPE Production & Facilities, February 1995
`
`IWS EXHIBIT 1058
`
`EX_1058_004
`
`

`

`Downloaded from http://onepetro.org/PO/article-pdf/10/01/7/2602443/spe-26525-pa.pdf/1 by Robert Durham on 20 September 2023
`
`HIT WITH NO FRACTURE
`Wellhead Pressure (psi)
`1,300
`1,200
`1,100
`1,000
`900
`
`b
`
`a = reflection from tubing mouth
`800
`b = r flection from bottomhole
`700
`600 uo----~::...:,j4:::.:.::::=::::.:...:.:.:.:.:.:...=.:=8:..:.:.:.:=-----:1-:!2
`
`Time (sec)
`
`HIT WITH OPEN FRACTURE
`Wellhead Pressure (psi)
`1,800
`
`c
`
`1,600
`
`1,400
`
`1,200 0
`
`c =combined reflection from
`tubing mouth and fracture
`
`4
`
`8
`
`Time (sec)
`
`1 8 0 0 . . , . - - - - - - - - - - - - - - ,
`A B
`
`HIT at 1680 psi WHP
`
`1700
`
`1600
`
`600
`
`550
`
`HIT at 570 psi WHP
`
`12
`
`o
`
`5
`
`10
`
`20
`15
`Time (sec)
`
`25
`
`30
`
`HIT WITH CLOSED FRACTURE
`Wellhead Pressure (psi)
`120
`
`100
`80
`60
`40
`a = reflection from tubing mouth
`20
`oLo-----.b..;=-.IJ;lfi4~QOJ[rQ[]1..lxillQlm8ne....----~12
`
`b
`
`Time (sec)
`
`Fig. 1o-Wytch Farm HIT traces.
`
`and these results indicate pulse attenuation in the wellbore of
`approximately 23% ofthe initial pulse as the pulse travels to the bot(cid:173)
`tom of the well and back. The reflection coefficient is therefore de(cid:173)
`creased to - 0.92. The hydraulic impedance of the wellbore is found
`from Eg. 4 as 8200 s/m2 and the fracture impedance is, from Eg. 12,
`342s/m2 . The fracture height is then obtained from Eg. 13 as 8.4m
`(with G = 8.5E9 = 0.2). The corresponding fracture width, obtained
`from Eg. 5, is 2.8mm. Calculation of fracture height and width for
`the HIT traces shown in Fig. 8 indicates that both fracture height and
`width decrease progressively as wellhead pressure is reduced.
`reaching about 7.2m and Imm at a well head pressure of 900psi. A
`slight increase in height is then found, although this is likely to be
`the result ofparameter uncertainty as fracture closure is approached.
`Fig. 9 shows a spinner trace for Magnus C7 indicating that the
`majority of the flow leaves the wellbore through a 40m interval.
`The HIT estimate offracture height is lower. This may be the result
`of poor wellbore fracture communication. Analysis of a pressure
`fall-off test on Magnus C7 indicates a skin of approximately - 1,
`which tends to support the view that the well is not well connected
`to a large fracture. However. it should be noted that HIT estimates
`of fracture height (most ofwhich have been made on deviated wells)
`are generally lower than expected, although they are often close to
`the heights indicated on spinner surveys.8
`
`HIT on a Newly Fractured Well. HITs performed, both on wells
`in the Magnus field and elsewhere, indicate that for wells which
`
`Fig. 11-South Ravenspurn HIT traces.
`
`have been injecting for several years, fractures do not appear to
`close fully when the wells are shut in. However, for newly fractured
`wells, fractures appear to close almost completely when the wells
`are shut-in. Fig. 10 shows three HITs from a Wytch Farm injector.
`The first trace shows the well before start up. The well was then
`deliberately fractured and the second trace shows the HIT response
`ofthe well with the fracture open. The well was then shut-in and the
`third HIT trace was obtained. This third trace is very similar to the
`first trace indicating that the fracture has closed. A step rate test. per(cid:173)
`formed concurrently on the well. shows very low injectivity beneath
`fracture opening and a clear change in the gradient of the plot at the
`same pressure as HIT indicates fracture opening.
`
`HIT During Hydraulic Fracturing. HIT was performed during
`the minifrac stage of a propped fracture treatment on South Raven(cid:173)
`spurn well A06, which has a near vertical trajectory across the reser(cid:173)
`voir. Fig. 11 shows two HIT traces obtained during a pressure fall(cid:173)
`off. one above fracture opening pressure and one below. The traces
`clearly show the effect of fracture closure.
`During the treatment a 5.5" diameter fracturing string was present
`in the well, which contains a 7" liner. The first reflection on the pres(cid:173)
`sure traces (A) therefore comes from the tubing mouth with the later
`reflections (B) from the fracture mouth.
`The fracture mouth response indicates two fracture zones and
`analysis (by matching the trace obtained from a simulation package
`with the field record) indicates that the upper zone has a height of
`just over 2m and the lower has a height of about Sm. The fracture
`length of the lower zone is estimated at just under 40m. Analysis of
`the pressure decline following shut-in from a higher injection pres(cid:173)
`sure gives a fracture face area of 27000m2, which is considerably
`greater than the HIT estimate. The discrepancy may be because the
`measurements are made at different excess pressures. It is interest(cid:173)
`ing to note, however, that a spinner log run after the hydraulic frac(cid:173)
`ture treatment shows that approximately 80% of the flow into the
`well is from two zones suggesting that the HIT result may be show(cid:173)
`ing real features of the flow profile.
`
`Summary. HIT has been shown to be a cheap and reliable method
`for detecting fractures and identifying fracture closure pressures.
`HIT appears to give reasonable estimates of fracture lengths, but
`
`SPE Production & Facilities. February 1995
`
`11
`
`IWS EXHIBIT 1058
`
`EX_1058_005
`
`

`

`Downloaded from http://onepetro.org/PO/article-pdf/10/01/7/2602443/spe-26525-pa.pdf/1 by Robert Durham on 20 September 2023
`
`IWS EXHIBIT 1058
`
`EX_1058_006
`
`