Professional Documents
Culture Documents
SPE 22559
This papar was prepared for presentation at the 661h Annual Technical Conference and Exhibition of the society of Petroleum Engineers held in Dallaa, TX, October S-9, 1991.
This papar waa selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(a). Comema of the paper,
es pfesenmd, have not been reviewed by the Society of Petroleum Engineers and are subiect 10correction by the author(a). The msterial, sa preaerded. does not necessarily raflect
any positicm of the Society of Petroleum Engineera, ita officere, or members. Papers presented at SPE meetings are subject 10publication reviaw by Editorial Committees of the sozialy
of Petroleum Engineers. Pemsisaionto tnpy is reetricfad to an abstract of not more than 309 words. Illustrations may not be copied. The abstracf should contain conspicuous acknowledgment
of where and by whom the paper is presented. Write Publications Manager, SPE, P.O. Box 633636, Richardson, TX 75663-3S36 U.S.A. Telex, 73G989 SPEDAL.
ABSTRACT the surface directly into the blowing well in the inwell kill.
When the inwell kill is impractical, a second option is to direc-
Three simple calculation techniques are developed and ex- tionally drill relief wells and iqject kill fluid into the formation
plored for determining the minimum dyne&ic-kill rate. ho of close to the blowing well. For either option, an accurate predic-
the techniques employ only single phase calculations and are tion of the kill rate is key in planning and pmptuing for a kill
independent of the reservoir inflow performance. Despite attempt. The magnitude of the kill rate may determine wheth-
these limitation, these simple methods provide a usefuI er an inwell kill is practical or whether a number of relief wells
means of bracketing the minimum flow rata necessary to kill must be drilled. The size and numtxx of relief wells, amount
a blowing well. and treatment ofkill fluid, sise and quantity ofpumping equip-
ment, and pressure rating considerations all depend to some
The third technique uses a simplified mechanistic model of degree on the magnitude of the required kill Gti.
multiphase flow to determine a most probable minimum kill
rate. This minimum kiU rate is expected to be somewhat con- Accurate prediction of the required kill rate is compro-
servative because of the assumptions of homogeneous no-slip mised by uncertainties. There axe uncertainties in our knowl-
two-phase flow. edge of the reservoir, the flow path geometry, and the fluid
properties of the formation fluids and the i@cted kill fluids.
Comparisons against limited field and laboratory data in- Furthermore, a significant uncertainty can be present in the
dicate that the models are physically consistent end perform pressure and flow rate predictions of multiphase fluids in
quite well. pipas2.
The basic idea of the multiphase solution is that for any Fig, 3 is acomparisonof system intake curves for three well
successful kill rate, the bottomhole pressure prediction must known multiphaee correlation and the no-slip model. All sys-
be greater than the sandface pressm% for any reservoir fluid tern curves were generated for a constant iqjection rate of 80
flow rate. In graphical terms, the system intake (wellbore hy- bpm, The Aziz et al.s correlation indicates that injecting water
draulic) curve must lie above or tangent to the Iflow Perform- at 80 bpm would not come close to killing the well. The Beggs
ante Relationship (IPR) curve. Fig. 2 show System inkke and Brill correlation predicts much higher pressure Imses
curves and IPR for the Arun data in Table 1. For a given IPR than the nc%dipmodel at in~asing gas rates and would gen-
the minimum kill rate would result in a system intake curve erally be expected to underpredict minimum kill rate. The Ha-
that lies just tangent to the IPR curve. gedorn and Brown model agrees very well with one of the
no-slip models at these high ratea. The two rto-slip:modelsdif-
The system intake curves are generated for various combi- fer in the diameter used to determine friction factor for a fully
nations of injection and blowout ratis by adding the pressure eccentric anm.due.This is discussed in more detail later, Very
losses due to hydrostatic head, friction, end acceleration to the different results maybe obtained for different multiphase flow
outlet pressure, models. Although the no-slip model is expectedtobe somewhat
conservative, by itself this model does not addresathe range of
pwf = pm + ~ Ap%+ Ap,+ Af). possible error in the calculation. In particular, determination
i-l ( ) ., ..,.. (1) of lower and upper limits would help to establish some degree
of confidence in the prediction of minimum kill ra@
where the flow path has been divided into m increments.
Bottomhole Iw3eure Match Solution (lower Ilmlt):
PH
A~=~ The bottomhole pressure match solution determines the
. . . . . . . . . . . . (2)
liquid injection rata necesseg to maintain the bottomhole
pressure equal to the reservoir pressureonce the well has been
Apf= fL ~
[ dh 2&
+x
144
n K. Pm V;
j=l
*&
1 . . . . . (3)
killed.
%=5 R..,.,........(7)
. . . .- ....-- ... .. .-. .
I
The accelerational pressuw drop is neglected for near in- For the zero-derivative condition,
compressible flow with a constant flow path diameter. Eqs. 2
8 PWf
and 3 maybe rewritten for a single phase liquid and substi- -. Oasqg-+O
tuted along with Eq. 7 inta Eq, 1. aqg . ...,.. . . (12)
pLv;
144
2& 1 . . . . . . . (8)
the numeratorcanbs set b zero and rearranged to solve for kill
rate,
apWf
K
2gcA2
[(P=+Pg)qL+ 2Pgqg1 - AP H+: The following empirical relationship for friction factor from
= Jainle was used for Newtonian fluids in this study because it
a% is acmrati, explicit and includes pipe roughness.
a Pg qgqmK
1 ~~+q
(11)
[
,=p.14-210g(:+3)~
(,5,
1
247
.
Because blowouts generally involve high flow rates, thi. We fist present the procedure to bracket the minimum kill
Reynolds numker maybe assumed to be infhitely large for rate solution, then discuss a simple method to estimate the
nonitarative hand calculations with little loss in accqracy minimum volume fraction for kill fluid below the POI. , .
The system intake curves in Fig. 2 were generated with the P GpLYu+pg YgQ. . . . . . . .. (18)
me
no-slip model and a ~ determined fi-ictionfactor. The results
of the two no-slip models in Fig. 3 show the magnitude of the Ibcompletethe calculation we need to estimate the volume
difference between calculating friction factor with dh and ~. fkactionof kill fluid below the POI. The lower limit kill rate will
be produced for 100% kill fluid below the POI, YLQ= 1. Assum-
NOtI-NWtOnlan Fluids:
ing 100% formation fluid below the POI, y~t =,O,generates a
The theological properties and a smooth pipe friction factor useful upper limit to the minimum kill rate when compared
for non-Ne@nian fluids can be determined fiwm procedures against the zero-derivative solution.
given in API 13d3.Unfortunately these procedure ignore any
effects of pipe roughness. At very large Reynolds numbers, the Zero.Derivative Solutton(upp?rlimit):
fticticii factor fora Newtonian fluid becomes a unique function
of relative roughness. Govier and Aziz 14have suggested that The determination of an upper limit kill rate is somewhat
the smooth pipe fiction factor fornon-Newtonian fluids can be more difficult for off-bottom injection. The zero-derivative so-
scaled up by multiplying the smath fiction factor for non- lution for off-bottom injection is still Eq. 13 but the vertical
Newtonian fluids by the ratio of fm.g~fsmooth for Ne*ni~ depth, total resistance coefficient and average gas density are
fluids at the same Reynolds number. determined with respect to flow path above the point of injec-
tion. Injection rates greater than or equal to the zero-deriva-
tive solution produce system intake curves that increase
OFF-BOllOM INJECTION monotonically with gas rate.
There are some applications when the kill fluid is tobeinjected The zero-derivative solution is an upper limit kill rate
some distance tkomthe bottom of the well, Fig. 4. Gillespie et when its system intake curve lies above the pressure at the
aL13studied the problem of off-bottim dynamic well control of point of injection that the reservoir can produce. This pressure
a sand-blinked gas well during workover operations to wash is csdculatid from the IPR bottomhole pressure minus the
out the sand bridges. A criterion for the countertlowof kill fluid presstire loss in the region below the POI. A conservative ap-
below the kill string was presented in Ref. 13,but there was no proximation for pressure losses in the lower region can be
discussion of the kill fluid volume fraction below the point of made by considering only the hydrostatic head loss due titfor-
injection (POI). Counterblowdoes not guarantee that kill fluid mation fluid below the POI.
will completely &spla@ all the reservoir fluid below the POI.
The injection rate that produces apressure at the point of
injection equtil to the bottomhole IPR pressure minus the
---
Z4U
SPE -_. .
22559 G. E. KOUBA.-, G. R. MCDOUGALL
--------- AND
---- . B. W.
--- -- SCHUMACHER
------ . . .. .. 5
i
hydrostatic head loss of the formation fluid up b the POI is de- The criteria for transition between slug and bubbly flow
termined horn Eq. 17 with 100% formation fluid below the was given by Ref. 19.
POL Therefore, the upper limit kill rate foroff-bottom ir@ction
is the greater of the zero-derivative solution or the bottomhole 1 . .
[1
u q/-~ 7
pressure match solution with formation fluid below the POL ()
Sg = 0,197 + 1.@ V$L
p:
Mo81Probabie Minimum Solution: . . . (22)
The thrust of the most probable minimum solution for otT- At equilibrium conditions there should be no net flow of ~n-
bottom injection is to better predict the volume fraction of kill jection fluid downward below the P(X Thus, vs~ = O.Combin-
fluid IM1OWthe POI. Accurate prediction of yu would mquim
ing Eq. 20 and 22 yields the minimum gas rate to sustain slug
counter-current floc@ing models beyond the scope of this pres- flow.
ent work We can, however, calculate some threshold formation
fluid flow rates below which a minimum value of liquid holdup
can be established.
%y4 =o.602~ m +
Fig, 5 is a flow pattern map based on a mechanistic ZT #
modelle for gas and liquid flowing upward in an anmdus for the [1 ,. .,. (23)
Arun well conditions. The solid lines represent transition
boundaries between flow patterns. The transition boundary This transition is baaedin parton the idea that bubbly flow
between annular and nonannukir flow was developed from a can no longer exist when the liquid holdup falls below about
force balance on a droplet of liquid in a gas stream. The transi- 0.75.
tion is marked by the minimwg velocity required to suspend
a liquid droplet. Therefore, by making a gross determination of flow pat-
tern, ranges of Iiquidholdup can beset, For the different values
of minimum liquid holdup, system intake curves for the bot-
H
u (PL_PJ + tomhole pressure can be generated from Eq. 16. Figure 6 shows
Sg = 1.593 different intake curves at a fixed injection rate. If the syetem
&2 .,. .,(19) intake curve for YU = Ointersects the IPR at a gas rate below
the annular flow traneitionboundary, then we are assured that
In a number of studiesl ~ 7*19,Eq. 19 has been found to ac- the lower region has passed inta slug flow and that the liquid
curately predict the annular flow transition. In a series of field holdup is at least 0.25. Similarly, if the system intake curve for
studies, Colemaa etat.g successfully used this criteria to deter- Y@ = 0.25 intersects the IPR below the slug to bubbIe transi-
mine when low pressure gas wells would begin liquid loading. tion rate, then we know that the liquid holdup is at least 0.75.
We can proceed in this simplified approach until satisfied that
The volumetric liquid flow rate is calculated from in-situ
the injection rata will force the transition fi-omannular to slug,
superficial gas vekwity by Eq. 20.
to bubbly and finally kill the blowout.
A p Sg Fewer calculations are required in simply substituting the
qgsc = 3.056 ~T
. ...,, . . (20) transition gas rate into the no-slip model, calculating bottom-
hole pressure and comparing this pressure against the sand-
Combining Eqs. 19 and 20 yields the transition criteria in face pressure ftom the IPR.
terms of standard volumetric flow rate.
We reiterate that this solution should be conservative, i.e.
overpredict HIIrate. The assumptions of n~slip flow above the
[1
PCN,negligible fiction below the POI, and minimum liquid
qgsc =4.87~~ i holdup for each flow pattern all serve to underpredict total
Pg2 pressure drop and thus overpredict kill ra@,,
. . . . . . (21)
Below this flow rate, liquid will begin to fall downward RESULTS
flooding the region below the POI.
It ia tiortumate that there are not many blowout data of the
Bamea 4 indicates that when blockage of the gas core by a
quality and completeness of the Arun data available for dis-
large supply of Iiquidin the film causes the transition fkoman-
playing the results of these dynamic kill models. When com-
nular to slug flow, the minimum liquid holdup required to form
pared against available laboratory and lielddaw however,the
slugs is about 0.25. This aetathe lower limit liquid holdup once
models and their individual components have performed quite
slug flow has been established.
well. The calculations for the Arun blowout are presented as an
I
I
example.
- .-
Z4V
6 ADVANCEMENTS IN DYNAMIC-KILL CALCULATIONS FOR BLOWOUT WELLS SPE 22559
The Arun data is given in Ihble 1. Eq, 9 is used to calculati gas density in this region was calculated assuming an average
the minimum flow rate to sustain the kill. The total flow resis- pressure of about 7000 psia. The upper limit to the minimum
tance, K, in Eq. 9 is comprised of the flow resistances fkomfric- kill rate from Eq. 17 was 105 bpm.
tion in the annulus, mud cross, elbows and exit, The friction
factor was calculated with Eq. 15 using ~ in the relative The range of the most probable minimum kill was signifi-
cantly reduced by considering the flow pattern transition
roughness and Reynolds number. The piping was assumed to
boundaries. The transition fkom annular to slug was deter-
be new commercial steel pipe with a roughness of 0.00015 ft.
mined to occur at about 18 MMscfld for near POI conditions,
Aftar two iterations on flow rate, the value of f was 0.01448, A
resistance coefficient of 496 was calculated for ~dh, the fric- SimilarlX the transition between slug and bubble flow ilom
tion resistance in the anmdus. The effective value of the other Eq. 23 WW.about 1 MMecf7d.The flow pattern transition crite-
resistances was about 2 giving a total K value of 498. ria at the point ofinjection and the no-slip model above the POI
indicated that even the lower limit kill rate of 96 bpm placed
After substituting into Eq. 9, with proper unite conversion, sufficient backpressure on the formation to reduce the gas rate
a value of 64 bpm for the lower limit of the minimum kill rate below the anm.darflow transition, Fig. 6. A value of YM= 0,25
was calculated. This is essentially the same value reported in
caused the system intake curve to intersect with the IPR at
Ref. 5. The calculations differ in that Blount5 used a much about 5 MMscfld. While this was not sufficient to guarantee
smoother 0.000054-ft roughness, an infinitely large Reynolds transition from slug to bubble flow using our simple procedure,
number and the hydraulic diameter in the calculation of fric-
it does indicate that the flow is well into the slug flow regime
tion factor in the annulus.
where the liquid holdup is probably high enough to move the
The zero-derivative solution was determined flom Eq. 13. system intake to intersect the IPR in bubbly flow. If the range
The average system pressure was estimated to be about half of liquid holdup is conservatively considered to be 0.75 to 0,25,
the reservoir pressure and yielded an average gas density of 9 Eq. 17 yields a range of most probable minimum kill rate be-
lb#t3. This gas density was used in Eq. 13 to calculate a zero- tween 98 and 103, bpm respectively.
derivative solution of just over 84 bprn,This indicates that for
the Arm well the minimum rate to sustain the kill is ideo ade-
CONCLUSIONS
quat.atu kill the well. Neglecting the gas density as in Eq, 14
resulted in a conservative upper limit of 97 bpm.
Simple procedures have been developed to predict the lower
As should be expected, the above minimum kill rate is in limit, the upper limit and the most probable minimum injec-
excellent agreement with the value of just over 82 bpm from tion rates of kill fluid needed to kill a blowing well.
the no-slip multiphaee model.
The lower limit is the minimum flowrate necessary to sus-
It was unknown if there was any flow up the drill pipe in tain the kill once the flow of fluids fimmthe reservoir has
the Arun blowout. Assuming no significant restrictions, other ceased.
.
than the ti-iction of the pipe, we can calculate the maximum The upper limit kill rate is determined as the flow rate
flow up the drill pipe. Calculating ffiction in the normal way which satisfies the zero-derivative condition and is suffi-
for pipe resulted in a resistance coefficient of 475 for the drill cient to maintain the kill. The zero-derivative Conditionin-
pipe. From Eq. 9 a value of 33 bpm was calculated as the maxi- sures that there is no more than one solution, i.e.,
mum expected flow in the drill pipe. The maximum flow intersection between IPR and system intake curves,
through both the drilI pipe and annulus is then 117bpm. It is A simple homogeneous no-slip model of the multiphase
interesting to note that the injection rate actually needed ta flow was shown to provide a-accurate and conse~ative
kill the Arun well was 114 bpm. This may indicate that less prediction of the most probable minimum kill rate.
fluid wee lost to the formation than predicted by Ref. 5.
The calculations for bracketing the minimum kill rate have
Calculations for off-bottom kill rates were performed on a been extended to the case of off-bottom injection of kill fluid.
well identical to the Arun well but with the point of injection Additional procedures have been presented to reduce the
2000 ft above the bottomhole. The values for H. and Ku are bracket around the minimum kill rate for off-bottom injection,
7540 ftand 394, respectively.The minimum kill rate to sustain A set of well-established flow-pattern transition criteria has
the kill is determined by assuming 100% kill fluid below the been used to &termine gas rate at which liquid will flow down
POI. Substituting into Eq. 17 reveals that the lowerlimit to the below the point of injection.
minimum kill rate is about 96 bpm in the amulus alone.
A~ change in pressure due to hydrostatic head, psi 1. Ansan, A. M,, Sylvester, N. D,, Shohan, O,, and Brill, J. P.:
A Comprehensive Mechanistic Model for Upward Two-
% production gas flow rate, ft3/sec Phase Flow in Wellbore, paper SPE 20630 presented at
~ gas flow at standard conditions, MMscf7d the 1990 SPE Annual ikchnical Conference and Exhibi-
tion, Dallas, Sept. 23-26.
~~ liquid injection rati, ft31sec
2, Ansari, A. M.: A Comprehensive Mechanistic Model for
h mixture rate; ~ = qL+ ~ Upward Ih-Phaee Flow in Wellbores, MS thesis, The
University of lldsa (1988).
T temperature, R
3. Whe Rheology of Oil-Well DrilEng Fluids, API Bulletin
v velocity, fWec 13D, 2nd cd., (May 15, 1985).
v% superficial gas velocity, fVSSC; W+g. WA 4. Barnea, D.: A Unified Model for Predicting Flow-Pattern
VSL superficial liquid velocity tise~ vsL. qJA Transition for the Whole Range of Pipe Inclinations, Int,
J. Multiphase Flow, 13(1987) 1-12.
Y~ liquid holdup; YL = 1 yg
5, Blount, E. M. andsoeiinah, E.: DynamicKill: Controlling
Yg gas holduw y~ = 1 yL Wdd Wells a New Way: WorZdOil(October 1981) 109-126.
z gas compressibility factor 6. Brill, J. P. and Beggs, H, D.: lluo-Phase Fiow in Pipes,
short course textbook, University of lklsa, Tulsa, OIL
Greek Third Edition (1978) 3-11 through 3-62,
pipe roughness, ft 7. Caetano, E.: Wpward Vertical Two-Phase Flow through
specific gravity an Annulus, PhD dissertation, University of Tulsa,hdsm
surface tension, dynesfcm OK (1986).
no-slip gas holduw ~ / (qL + C@ 8. Clark, A. R. and Perkins, T. K.: Wellbmw and Near-Sur-
fice Hydraulic of a Blown-Out Oil WellVJ. Pet. ZkCh.(No-
no-slip liquid holdup; w/ (qL + @
vember 1981) 2181-2188.
density, lb#t9
9. Coleman, S. B., Clay, H.-B., McCurdy, D, G., and Norris,
PL-Pg
H. L.: A New Laok at Prediction Gas-Well Load-UpT J.
viscosity, cp Pet. 2kch.(March 1991) 329-33.
StMCripts 10. Dodge, N. A.: Friction Lasses in Annular Flow, ASME
63-WA-11, Winter Annual Meeting, Philadelphia, PA
a acosleration (Nov. 17-22, 1863).
c casing, circulating
11. Dodge, D. W. and Metzner, A. B.: lhrbulent Flow of Non-
Newtonian Systems~AIChE J, 5 (June 1959) 189-204.
251
8
252
..
~PE 22559
WAY 1
09fa for Awn Lflawout
7100
Rti PW4UMW
@d
ROWIVWIanqwahm F 234
sbmpil raw, MMacfld 370
0.4167 0.2917
--,ff
LMHIP@lD,fi o.35e3 0.2493
AIWWIV8
d, n 0.2946
NKUIIUS
~. n 0.5784
Anwlus area! n? 0.20096
0.00015
P@ roughness,H
Mea6uradd@ffI, if 10210 10900
TW Vwf!ealdacth,ff 9650 9540
KUIfkJHdetWfY,L%I@ 61
8.155 . ..>
(fraahwaferaf lSOF,3#gal)
_alul injeSbn kill We, @m 114
Fig. 1 Schemtic of bfowoutwell showingtwo possibleoptionsof
Dda adknabdt?/mllfwrs dynerniceilykiilingthe well, imveiiand reliefwell injection,
lo-
8.
7-
= 6
z
~5
F
3-
2-
1-
0
0 m m 3W w
~ (MMeof/d)
Zes
4--- %
%+ qL h .
i
Fig.3
2
1-
+ no slip&~
+ f10-S@ & dh
A EtqgS 6 Brill
x Hagadorn
Azizatal.
q (kiii%ofld)
Ffg,4
1 Pointof injectio~
lM. D18PER8ED
B~BBLE
10=
g, BUBBLY
e
u
.ANNUIJ 1:
0.100
: =x
O.oto
~
2s4