Documente Academic
Documente Profesional
Documente Cultură
of
Potroloum En@rseers
U.S. Department
of Energy
SPE/DOE 13905
SPE Member
This paper waa presented at the SPE/DOE 1985 Low PermeabilityGas Resewoirs held in Denver, Colorado, May 19-22, 1985. The material is subjectto
correction by the author. Permission to copy is restrictedto an abstrast of not more than 300 words. Write SPE, P.O. Sox SZZ886, Richardson, Texes
75WW383S. Telex 7300s9 S% DAL.
ABSTRACT
‘On account of their highly non-tiewtoni
an
fluid characteristics,the common fracturingfluid
Experimentalresults of the settling of
possess completelydifferent propertiesunder
spherical particlesin flowing non-Newtonian
shear and under stagnant conditions.In the
fracturingfluids show that Stokes Law based on
calculationof proppantsettling velocities,two
‘Power Law! viscositiesis insufficientto predict distinct shear regimes must therefore be
particlefall rates in both flowing and quiescent distinguished:
fluids. 1. During the hydraulicfracturingtreatment,when
the fluid is being pumped.
In a stagnant fluid the experimental
2. During closure of the fracture followingthe
settling velocitiesare more than an order of treatment,when the fluid is essentially
magnitude higher than those calculated,while in a
stagnant.
flowing fluid, settling is lower than that
calculated.These phenomena can be explained by In both cases knowledgeof the settling
extending the ~PowerLaw! model with a zero shear
velocity of a single sphere is the first step in
viscosityand by assuming an anisotropicviscosity
understandingof the complex transport process
in a flowing fluid.
leading to the final proppant distributionin the
fracture.
Anisotropyin the viscosityonly beS?mes
important above shear rates of, say, .25s , and
There are few publishedexperimentalFes.iilts
so will not play a role in the majority of
of the settling of proppant in a quiescentor a
fracturingtreatmentswhere average shear rates in
flowing fluid. The build-upof a proppant bank
the fracture will be below this value.
from a non-Newtonianfluid in a vertic~l slot flow
model w s studied by Schols and Visser and
3
Babcock . Several other’sstudied the settling of
1. INTRODUCTION single particlesu der shear in concentric
4,5
cylin er devices ,a moving belt par llel plate
In a hydraulicfracturingtreatment,a ~ ?
model or in vertical slot flow model .
fracture is created from the wellbore by rupturing
the formationat high pressure by means of a Each of the authors report findings which
fracturingfluid. A propping material, carried by deviate fran Stoke’s Law settling velocitiesin
the fracturingfluid, is placed in the induced sheared fluids as well as stagnant fluids, but no
fracture channel to prevent the fracture fran good explanationwas found for this anomalous
closing after the fluid pressure has been behaviour.
released.The productivityimprovementis mainly
determinedby the propped dimensionsof the This paper provides an explanationfor such
fracture, which in turn are largely controlledby behaviourby using a descriptionof the
the settling velocity of the proppant in the frac theologicalbehaviourwhich accounts for the zero
fluid. A high settling velocity will result in the
shear viscosityof the fluid, and by Introducing
formatio~ of a proppant bank at the bottom of the the concept that the resistanceto motion across
fracture . A very low settling velocitywill the flow is differentfreesthe resistanceto
permit the proppant to remain in suspensio~ motion in the directionof flow. For conceptual
distributedover the total fracture height . ease, this is best describedby the term
‘anisotropicapparent viscosity . However one must
realise that the term viscositycan not be
uniquely defined for a non-Newtonianfluid.
---
.
2. STATIC SETTLING
For9’PowerLaw’ fluids, Reynolds number is defined
The ‘powerlaw’ model, commonly used to as
describe frac fluid rheology, is considered
adequate for descrl’bingthe p~~pe?tie~Gf imst Re=+
fluids under ehear, but, as is shown below; it
does not accuratelydescribe settling of the
proppant in a stagnant fluid. which transformseq. (1) into
Methods for predictingsettling velocities I/n
dpg-pl)d
of propping4agentshave been publishedby Daneshy8 v~={ ,8K } d (6)
and Novotny . Both methods employ a form of
Stokes’ law which states that the drag coefficient
acting on a slowly falling spare is given by: The settling velocityhas now becom~l$n{fiction
of n and K, ~nd 1S Proportionalto d
instead of d ss in the case of Newtonian fluids.
The validity of equation (8) was tested These problems can be avoided by using a
experimentally.A cylindricalvessel (diameterD) parallel plate model and accepting the fact that
was filled with test fluid, and glass or steel slow settling,such as occurs in cross linked
balls (diameterd, density p) were allowed to f~l gels, cannot be studied due to the finite length
along the axis of the cylinder and the terminal of the model.
velocityof the fall, V , measured. In this
experiment,spheres with diameters of 1; 2; 4; 6; The overall shear imposed on a particle
8; 10 and 15 mm were used with cylindershaving falling in a flowing (sheared)non-Newtonianfluid
diameters of 49.5; 35.3; 25.9 and 11.4 mm. As the is very difficultto calculatebecause the stream
measurementswere taken in a liquid with a finite function is notl~nownfor such a syst~. stokes’
volume, a correctionfor the wall effect has to be stream function can be used to obtain a surface
made. The infinite bath fall velocity then becomes average shear rate acting on a sphere falling in a
quiescentNewtonian fluid (AppendixA).
‘t
V. = ~. The correctionfactor, F, was determined
This results in an average shear rate
in the same manner as for a Newtonian fluid by 2vt ‘t
plotting the measured settling rates against the
ratio of the sphere diameter to the diameter of <Y>’= —. In the literaturevalues frcm <~> = —
d d
the cylindricalcontainer,for all sphere-cylinder
combinations.Vm is determinedby extrapolationto 1. 33V
to <Y> = — for power law fluids are found
d n2d
zero -
D. In Fig. 2, the results are comparedwith a
(n is the power law index), dePendingon the
theoreticalcorrectionfactor for a N~y~onian stre~~unCtiOn used. An average shear rate
fluid, proposed by Haberman and Sayre . The good
agreement between theory and experimentalresults CT> = ; gives the best results in fitting
confirms that at low shear rates the fracturing
fluid exhibits Newtonian behaviour (Fig. l). experimentaldata and is in agreementwith eq. 6
and 7.
The correctedexperimentalsettling
velocities can be comparedwith the fell rates If one assumes the proppant particlesto
calculatedwith equation (8) and with eQIaUOII travel with the flowing fluid and that the
(6). This comparisonis depicted in Fig. 3. It is horizontalstream function generatedby the
clear that the use of a ‘power-law’model flowing fluid does not influencethe vertical
(equation6) to calculatesettling velocities, stream function generatedby the falling particle,
the overall shear rate acting on a proppant
P~~*Jc~s ~a~ge~ ~het czc & ~Qpe than an Order of
magnitude too small. This may be particularly particle is the vector sum of the ~hea~ rate *Je
importantwhen a long fracture-closingtime is to p;oppantsettling, (y) and the shear rate
expected (as is the case in the tight gas imposed by the fluid mo!?
ion at the location of the
reservoirs),since a correct estimateof the falling particle.This assumptionis valid if one
proppant settling-rateis then essentialto ccmponentof the shear rate is small with respect
calculate the ultimate propped fracture height and to the other.
thickness.
Using this overall shear rate, the viscosity
is calculated,and used in Stokes’ Law to
3. DYNAMIC SETTLING calculatethe settling rate of a particle in a
Plowing fluid.
During closure of the fracture,the
fracfluid is static and the settling process is This ie the basis of Novotny’s descriptions
governed by the Newtonian part of the viscosity of the settlingmechanism in flowing Power-Law
curve. During pumping, the fluid is subJectedto fluids. However, this approachdoes not take the
shear and settling is controlledby the non- existenceof a zero-shearviscosityinto account,
Newtonian part of the flow curve which should which causes the model to fail at low shear rates.
result in faster settling. It also assumes the viscosityto be independentof
the directionwith respect to the flow i.e. the
Sedimentationof proppant in sheared fluids fluid is isotropic.It can be envisagedthat in a
has been studied by several workers using flowing polymer solution shear will induce an
different experimentaldevices,varying frcxnslot anisotropyin the fluid by alignment or stretching
flow model to rotating concentriccylinders. of the long chained polymermolecules.This
Rotating, concentriccylinder devices have the orientationwill be opposed by Brownianmovement,
advantageof an infiniteslot length, but the which tends to disorder the system in such a way
disadvantagethat centrifugalforces drive the that at each higher shear rate an equilibria
P?’OppaEtwrti~l~~ Eoward8 the outer cylinder aat.hl
-“”..”- feha=
..J..-” with
“. . . . -a vfldlleed
. ------ ~~~a~~nt VISCOSit~
wall, which, because of the complicatedshear along the flow direction,but not necessarily
profile in the annulue,makes interpretationof perpendicularto the flow direction.
the data a hazardous exercise.Another problem
with suc a device is the occurrenceof Taylor
vortices5 when the the inner cylinderrotates.
13905 L.P. ROODHART 1.
.—
5 PROPPANT SETTLING IN NON-NEWTONIANFRACTURINGFLUIDS SPE 13905
CONCLUSIONS
~2
1 3s
The Power Law model canmonlyused to 22
‘=vRsine[-G+T-2 —I
describefrac fluid behaviouris inaccuratewhen
used to calculateproppant settling, either in a
static situation during fracture closure or in a r
dynamic situation during pumping. with c = -
R and R the sphere radius, the velocity
Vo. o
—
L.P. ROODH.ART
SPE 13095
Gelling agent B. Pm T
‘LUID o
IUMBER concentration
kg/m3 Pas Pas Pa
HEC
4.8 ().44 0.14 1.8
I
7.2 1.6 0.35 7
11
HPG
3.6 0.22 0.07 0.6
I
4.8 0.40 0.16 0.7
II
7.2 3.4 0.7 7
111
v,
APPARENT
APPARENT VISCOSITY Po, ymer ,0”., ”,,0!!0”
VISCOSITY
(POISE) . ,2 0 kQ/m’
( Pas)
1000 A 96
ioo
—. FIT WITH EQUATION7( EXTENDED POWER LAW)
: ::
36
+
xsxxs= ExPERIMENTAL CURVES
“j,,;LAO
The
0
sol!d
0! 02
curve IS
03
occord!ng
04
to
05
Habermon’s
06 07
exact
+—
theory
Fig, l-tirent wsi~S. shear rare fmplymer tiut#ons mthprownnm mllst&!n Tabk 1.
Fim 2-Wall correction factw for a cyimdncal V-I.
MEASURED sETTLINGS
VELOCITY (m/s)
,().1
I ‘/”
t x :x X** ●
x *x ●*
x x xx ,x ‘.** ●
x ●
x
xx ●
x x ●/
I
xx
40-3 x
●
x
x
PoWER LAW
ExTENTEDPDWER LAW
{0. 4
I
100% AGREEMENT
w.__.
cl- L,-”d.l ...—
Klc&l,
{0-2 {o-i
,(3-4 ,(3-3
+0-5
‘“’s’O1
/
‘L’Z=T....
K-&
F
● 048%HEc
x06%
❑ 072%
SC2LUTIUN
1,
0.1 i
f, ~-i
10 !00
-----
lUUU
1
F*. s-mm hmzcmtd (WU82) md VOltkA (Im2) .wIllpcnanta 01tiwaF@arOntvlsms@ VS.Sh+a ,*O i. a
HEC Sc4diin.
‘a”-:
i .0
—- II
--&_
--~
o i-
I
L..
.
i
10.1
0.0{ {0 100 {000
O.i t.
+. ~.i
FRAC.
WIDTH
mm.
AFTER TOTAL VoLUME
20 - IS PuMPED
~-
10
1 f~ AFTER PAD VOLuME t Izzfi
I. J//,”
WA
IS PuMPED
\ MAIN sETTLING “f
I // \ AREA
IY 0 2 4
1
-4 -2
LOCATION lNFRACTURE, mm
MEAsURED sETTLING
vELC)CITY(mm/s)
/
10 -
. POWER LAW EQ 4
* EXTENOED POWER LAW EQ 7
i /
z ❑ EQ9
/
(a) ●
10 ✎
50
I t
. .“
7X 0.1 -
30
(s-1)
20
1
{0 I
1
{ 40
0.4
CALCULATED SETTLING VELOCITY ( mmis)
Fb. 11-w- tuvmenrmawti ●ti mbtied sealhw
.mlociMs
Ofat.mm@we inanow.
,.E0.72%W= ~~~