Professional Documents
Culture Documents
Prediction of Vertical Hydraulic Fracture Migration Using Compressional and Shear Wave Slowness-1 PDF
Prediction of Vertical Hydraulic Fracture Migration Using Compressional and Shear Wave Slowness-1 PDF
of
Petroleum Engineers
U.S. Oepartmont
of Energy
SPEIDOE 13895
This paper wee presented at the SpE/DOE 19S5 Low PermeabilityGas Reaawoira held in Denvar, Colorado, May 19-22,1985. The material is subjastto
correction by the author, Permiaaionto copy is restrktad to an abstract of not more than S00 worda. Write SPE, P.O. Sox SSSSS6,R)chardeon,Tesea
7S0S3-SSS6.Telex 7S0989 SPE DAL.
questions concerning fracture migration consists Much research, both theoretical and applied,
primarily of in-situ stress, tensile strength, and has been conducted in recent years toward greater
elastic constants of the rock material in and understanding and control of fracturing
around the forrnat.ion Qf i~tey~~t. t.restriierlt,s.~~thougn severai general three-
dimensional computer-based models have been
This paper describes the use of full waveform developed, their applicationhas often been limited
data from a sonic wireline tool to determine the because of poor input data. Consequently, many
relative stress distribution and the resultant rule~f-thumb schemes are often employed for local
induced hydraulic fracture height. Compressional areas. When the proper input data are available,
and ahear wave slowness, derived from the sonic most of the more sophisticatedmodels can predict
waveforms, are used to calculate the dynamic the size and shape of a created fracture.
elastic rock properties. A transverselyisotropic Therefore,powerful tools exist in the industry for
model is used to compute the in-situ stress from efficient fracture treatment design but have been
the elastic properties. Advantages of the use of underutilized for lack of sufficient field data.
wireline meaaured data are discussed, as are the The consensus among investigators is that the
limitations of the technique. Final evaluation of information most needed for making realistic
~he technique is shown through the comparison of fracture geometry design decisions consists of the
predicted and poststimulation measured vertical elastic parameters of the rook, the in-situ stress
fracture height. TWO field cases are presented to conditions,and the created fractureheight.
illustratethe technique.
In general, this information has been
IN’r$oDucTIoll unavailable except on a relatively few research
type operations. Therefore,data from a few wells
Uith the recent advent of uncertain prices for oil are being extrapolated for uae over large regions
and gas, the importance of efficiently developing ao that predictions from even the most
hydrocarbon resources has increased sophisticatedmodels often have a large degree of
significantly. Many costs are relatively fixed and uncertainty. Usually, the greatest uncertainty
cannot be greatly impacted by improved about the created fracture geometry arises from the
technology. However, new technology can make estimation of the vertical migration or height of
significant contributions in hydraulic fracturing the fracture with changes in treatment pressure and
operation. Often, a hydraulic fracturing pumping conditions. A maJority of hydraulic
treatment not only represents a large fraction of fracturing models require a realistic fracture
initial well costs, but also determines the height as an input.1 Equations in fracture models
economic viability of a particular well or field. also require.elast,icconstants (Poissontsratio and
.lZn
PREDICTIONOF VERTICAL HYDRAULICFRACTURE MIGRATION
2 USING COMPRESSIONALAND SHEAR WAVE SLOWNESS SPE 13895
moduli) in order to calculate fracture length and newer sonic logging tools have receiver spacings
width. With information from wireline logs such that the compressional and shear waves are
(compressionaland shear wave travel time), it is sufficiently spread out in time to make routine
now possible to quantify the changes in vertical detection of the shear wave possible (Fig. 1).
fracture migration at the wellbore resul!-ingfrom Sophisticatedcomputer analysis of these digitized
changes in treatment pressure, elastic values of wavetrains has provided the capability of rapidly
formation rock, and boundary lithology. This paper extracting a shear slowness curve.10 After first
shows the use of combining sonic waveform analysis motion is detected, automatic windowing of the
with bulk volume analysis and treatment fluid waveform allows frequency domain operationa to
density (1) to yield an accurate picture of the in- locate the shear wave arrival. This technique
situ stress distribution in and around the eliminates the need for an analyst to continuously
b..,A..-e-...l.-..
,,JU, k--.-a
UbaaUu,l-u=ar
Lng fOrm~tiGF1, Urlu
‘-” (2) t.cl pmlduee decide visuaiiy what constitutes the first shear
an accurate prediction of fracture migration arrival from the various other modes of
behavior in relation to bottomhole treatment propagation. Fig. 2 presents an example of a
pressure. Variable Density~ log display of a typical section
Qf QpeRhQle Yz$?efe?’fi
ectz, Showifigthe ccxapiitier-
COHPRESSIOWALAWD SHEAR SLWUESS located first motion and estimated compressional
and shear arrivala before final processing. With
The heart of this procedure is based on the proper measured values of bulk density and compressional
utilization of information contained in the and shear slowness available, it is then a
wavetrain of an acoustic wave traveling through a straightforward process to calculate the in-situ
section of the formation. In traveling through a stress, sx, and the elastic constants which
section of rock, an acoustic pulse deforms the rock characterize the pay zone and boundary Iithology.
and, in turn, has its propagation characteristics Appendix A details the derivation of the in-situ
altered by the rock. Measurementsof the slowness stress.
of the compressionalwave yield information about
the reaction of the rock to a longitudinal FRACTURE HEIGHT
stress. Likewise, the slowness of the shear wave
is a measure of the reaction of the rack tm a Fiecent developments in tne modeling of
stress in the transverse direction. By combining hydraulic fracture geometry have shown that the
this data with bulk density it is then a migration of hydraulic fractures is primarily
straightforward process to calculate Pofsson~s dependent upon in-situ stress distribution and
ratio and other elastic constants of the rock. The treatment pressure. Simonson et al.2 has presented
dynamic definitionof poissonfsratio ia8 fracture equations that relate wellbore injection
pressure to in-situ stress distribution, material
2 ‘properties,and fracture height ❑igration. The
Y= [0.5 (vc/v5)2- l]/[(vc/vg) - 1] ....(1)
successful use of the Simonson et al. concept to
calculate variable height at the wellbore has been
and the dynamic definitionof Young*s mdulus is reported by Voegle et al.3 and Settari.4 Ahmed5
has shown how this concept can be applied to the
hydraulic fracturing models of Geertsma and De
E = Pbv: [(3V:..4v:)/(v: - v:)] x 2.15 X 108 (2) Klerk6 and Perkins and Kern~ for design of
optimized hydraulic fracture treatments. As
depicted in Fig. 3, this fracture migration model
where
has been modified to include fracturing fluid
gravity effects and can be summarized in the
‘JC. compressionalslowness followingequations:
V5 = shear slowness
Pb = bulk density of rock
C.,..b.maour-emcrl
““s,,
-A-- . . . —-.-k-
have been a IA standard
laboratory procedure for several years. Houevere
until recently it has been impossible outside the
laboratory to measure accurately and consistently
the slowness of the shear wave. Attempts to +C2 “
(Ub - Ua) Cos (:)1
estimate the shear slowness based on lithology and u
compressional slowness have met with only limited
.sllnna!ia
-.-..”-”-
. The ~a%p?e=si~iial ii=vetravels through a
formation at higher velocity than the shear wave. ’03 ‘f+ ‘hu-o”5h) ““”””””””*””””(3)
For proper detection of the shear arrival, the
travel path of the wave must be sufficientlylong
to allow the earlier arriving compressionalnave to
AP =PW-PO=C1 [Kic(+&)
dampen before the first shear arrival. Many of the
d
● Mark of schlumberger
SPE 13895 B. M. Newberry, R. F. Nelson, and U. Ahmed 3
m+
-.
PREDICTIONOF VERTICAL HYDRAULICFRACTURE MIGRATION
4 USING COMPRESSIONALAND SHEAR WAVE SLOWNESS SPE 13895
that can be applied and still remain within those to only the pay intervals, give no information
bounds. This informationis immediatelyuseful in about the location and quality of the barriers.
defining where barriers are located and quantifies Economically, the cost of routinely coring long
the containment potential of each barrier. It iS sections is usually prohibitive. Likewise,to gain
possible to quickly determine if sufficient minifacture information about the prospective
containment exists for a particular fracturing barriers would require perforating, testing, and
treatment or to assess the risk of fracturing into then squeezing off each tested interval. Such a
nearby water zones. The application of this procedure of actually fracturingthe barriers would
t.mr-hnintna +- u-ll- COil~aiiiiiig multiple perforated be an expensive and unwise method. This analysis
--- . . ..-Y”” WV -=-A-
intervalsis describedin another paper.~ is based on dynamically determined elastic
constants, while hydraulic fracturingis considered
h-situ stress in the perforated interval is a quasistatic process. The dynamic constants tend
treated as a single, lumped value. For the to be consistently higher than static measured
boundary lithologies, in-situ stress values are values. Therefore, there should be a consistent
used every 2 ft (0.6 m) as they are encounteredin offset to the calculations if static values are
❑oving up and down from the perforated interval. used instead of dyanmic data. However, the
The contribution of hydrostatic pressure as the fractura height prediction model uses only stress
fracture migrates out of zone is included in the differences and not absolute values ——— with the
calculation by using an input value of borehole result that any consistent shift has little effect
fluid density. on the calculations. Should core or minifracture
data be available, it is a simple procedure to
FIELD EXAHPLES include this information,in effect, to calibrate
the output to yield accurate absolute pressures.
The example illustratedin Fig. 4 shows the results
of the use of this technique on a section of the CWCLUSIONS
Cotton Valley formation in east Texas. In this
zone, the calculationsindicate that there is very
The QS~ Of ‘iy~li!?~ -“”.”..
‘nn*qa++C tiat,a ifi tjCjfij~~~~i~~
good containment.for net treatment pressures up to with proper computer modeling provides an accurate
about 900 psi. By the time the pressure rises to and cost-affective means of obtaining important
1050 psi, the height will have almost doubled, with
information needed for designing a hydraulic
much of the fracture treatment going into the side
fracture treatment; i. e., fracture height,
beds. For net pressures greater than about 1200
Poissonts ratio, and Youngfs modulus. Such data
psi, the situation worsens and most additional
can be readily used by operators and service
fracture fluid goes into the adjacent shales. The
companies as input to their computerized fracture
perforated interval was 12021 to 12016 ft (3665 to
design programs.
3662 m) and the maximum net treatment pressur~ vzs
approximately 900 psi. For this net pressure, the
MOHEMCLATURE
predicted total height of the created fracture was
11 ft (~=s ~) f~Q~ ~~n~u,GVG7 to 1~~1~ f~ (~&~ to
After-fracturemonitor logs indicated Vc compressionalslowness
3661.5 m).
the fracture extended from 12024 to 12012 ft (3665 ‘JS shear slowness
~Q 3&j~.3 fi) fO? ~ tGtd h~ight ~ %’uzglS md”ul’us
of 12 f~ (~.~ M).
bulk density of rock
‘b
fracturingflul& density
The second example (Fig. 5) demonstrates this
tehnique on a section of the Blanco formation in 2P net pressure above initial fracture
extensionpressure
Ojito County, New Mexico. Perforationswere from
Pw wellbore treatment pressure during
8068 to 8081 ft (2459 to 2463 m) and from 8110 to fracturing
8135 ft (2472 to 248o m). For this well the
P. pressure required for fracture
maximum net bottomhole treatment pressure was
extensionbeyond initial breakdown
approximately 1000 psi. The predicted total
hu fracture height migration into upper
fracture height was 118 ft (36 m) from 8040 to 8158
barrier
ft (2451 to 2487 m). The after-fracturegamma ray
fracture height migration int~ huer
log indicated the created frsc%!!reextefitieti frm
barrier
8039 to 8159 ft (2450 to 2487 m) for a total height
h pay zone gross height
of 120 ft (37 m).
ht total fractureheight
kic critical stress intensityfactor
COHPARIS(NOF HETSODS
a pay zone total matrix stress
a
The primary benefit, from a scientific and (fracturegradient pressure)
practical viewpoint, of using wireline data for upper barrier total matrix stress
ab
such an analysis is that continuous data are (fracturegradient pressure)
obtained. Measurements of cores or minifracture u lower barrier total matrix stress
c
results normally yield only isolated, sporadic (fracturegradient pressure)
values. Also, minifracturedata, normally confined C2* C3 = numericalconstants
SPE 13895 B. H. Newberry, R. F. Nelson, and U. Ahmed 5
s Y Clp) +p ..............(A-9)
s =0 + ap ..........................(A-4) 1
x ‘7(sz-
x x
SubstitutingA-3 in A-4 This model has provided results which are in close
agreementwith actual field data.
s Y + ap .......9..*.... ...(A-5)
x ‘~~z
1
● ●
The calculationof fracture gradient in an area
that cannot be classifiedas a tectonicallyrelaxed
basin is not straightforward. In basins containing
Rearranging Eq. A-2 and substituting the
numerous thrust faults, in areas near mountain
vertical matrix stress gradient (uz) into Eq. A-5, ranges, or in basins which have been uplifted over
the equationwould become geologic time, externally generated horizontal
stresses can affect fracturing pressures. Eq. A-9
should be rewrittento include such cases
s Y Up) + ap ...........(A-6)
1
x ‘7(sz-
Eq. A-6 shows the minimum horizontal stress s~ =*%- ap) + p + P+=- ...o.o(A-1O)
.=-
relationship for a homogeneous, isotropic, elastic
material. In hard-rock areas, this model does not
seem to adequately describe existing stress Eq. A-10 is a more general form of th fracture
conditions. Another model of in-situ conditionsis gradient equation. The first two terms of the
to assume the presence of microcracks in the equation express the portion of the fracture
rock. Such microcracks would tend to be aligned gradient that is caused by the vertical stress
perpendicular to the least principal stress. gradient and the elastic properties of the rock.
Physically, the result of using such a model is The third term, ptec, includes the effects of
externally generated horizontal forces. Normally,
that in the direction of least principalstress, it
is appropriateto use Ptec is determinedempiricallyby using field data.
[ FLUID
f
,-
SHEAR
RECEIVERS I
r
C~MPRESSION~L
‘, ~11111 ‘
I
................
%
11!
1’
. t II U“ll
I I -
---------------
111 11111111
~
ii TRANSMITTERS
‘“q%Rg’s2!!0:Py
rre.alm.n d WI). d{sln butl.n w,th tlm memo,.
BUI&V(’WI\~ME
I
Fmd”re height migrotrm at
the .ellb.re d.e t. i..m,., ;“ created fr.ci.m he~ht.
,. tr.otme.t pre.s.r.
the dew r, pre#swe
.b. am
100 z o
‘l----
I
s- Continuous I
l— Inmw”
(m)
4
L I
-------
--------------
-_-----_:----
-------
-----
-----
-------
-------
-------
-------
-------
---------
-----------
-------------
-----------
-------
-----
-------
-------
-------
.:-=-=-
-------
-------
-------
2100 _l
Figure4—Typical FracHite log presentation.
~ 13895
FRAC HEIGHT
NET P%ESSURE
,f at
miydii
Qy’’!’%.~J&y~Ry
Tmd-mn vd.me di.t.b.tion wth the ikr.ase
BULKJLKV~l#fE
1..to incm.mes in meow fraci”rs Iw,qht .
Pm.w.
,. pressure.
ob.w 100 z c
I
.= 13895