Sei sulla pagina 1di 7

lADC/SPE 35120

Comparison of Steady State and Transient Analysis Dynamic Kill


Models for Prediction of Pumping Requirements
L,W. Abel,

IWC Engineering

Co., and D.W. Shackelford,

Halliburton

Energy Serwces

Inc

SPE Members
IADC Member
@ @ylklflt ! S9S, lAIXXPE Drillhg ContWame
This paper was pracwad fcf waamtatm_I at ma 19S6 IAOCLSPE DriltktgC@eranca
Or!eens, Lcwdana, 12.15 March 19SS.

pure friction, 2) steady state two-phase and 3) transient two-phase models. For each analysis, the same
reservoir and well geometry were used so that the
comparisons would yield meaningful results. The definitions of each method are as follows:

hakj n New

ThM papar was aalactad fcf preaanfal~ by the IALXYSPE Pt ram Cornm#faa _
rawaw of
rnf~afii
confanetf in m abstracf sub+m
iffad ~ the *.MIo%). C@tWItS d the IMIXI as wSHIM, have nOf *
rtiWd
tY lfw SCCkfY 01 WOIWIM E~ItKW8
or m. lnlOMIiltid
&c&Ikm of C7W?g Ccntradcfs and are aubjscf 10 cmectmn
aantad, does no! nacesaati
rdlacf any Foshn
d the 1AO%:?&%%oJkv%%%R
PaWrap,eaenttialthelAkSPEmW!mgarea.w
to ~fkm
review by R&id
.cmmii.
ma of lha IAOC and SPE. Permiawcm 10 copy la reslrzfad to an abstract of not more fhan X0
words Iflualrakma may no( be ccpml. The abatracf ahou!d ccotarn ccmpcwua admmdadgmant of
SPE, PO. Box S3339S6, f7kfMK)
where and
whom the paper was praaantad Wrte
am, lX7&3S3S
USA

Librarian,

Pure friction-The
friction losses generated
by fluid flow plus hydrostatic pressure combine
to exert a pressure against the reservoir. The
design criteria is that the model estimates the
kill rate that will produce a pressure at the
reservoir that is greater than or equal to the
static reservoir pressure.

Steady state two-phaseThe


friction resulting from a steady state two-phase flow is such
that the compressibility and other important
non-Newtonian behavior of the fluid mix is accounted for in estimating the reservoir IPR
(Inflow Performance Relationship) effects in determining the minimum kill rate requirements

Transient
two-phaseThis
model uses twophase calculations for fictional
losses and
phase behavior, but also estimates the time
dependency relationships of the reservoir (IPR,
changing fluid compositions, etc.) and the fluids in the blowout flow path. This model is
useful
in determining
real-time
pumping
schedules for kill operations,

ABSTRACT
The purpose of this paper is to compare three methods
of calculating dynamic kill requirements for a blowout
well. The use of computers, especially desktop PCs,
has made complex analytical solutions more readily
available in solving engineering problems. The power
and capability of the computer has increased dramaticzdly in recent years, as has the analytical complexity
and effectiveness of engineering solutions. This paper
compares three methods that have been used successfully in the field to dynamically kill blowout wells, either from relief wells or through a tubular conduit in
the well. The methods to be compared and discussed
in this paper are:

Pure friction dynamic kill

Steady state two-phase

Transient two-phase flow analysis

flow analysis

PURE FRICTION

To make the comparisons useful to the reader,


each of these analytical methods has been applied to
the same type of well with identical reservoir conditions and characteristics. The conclusions show that
the kill requirements
can be estimated
by using
simplified models, but to more accurately estimate
pumping schedules (i.e. volumes and pumping times) a
complex model and analysis is required. A tabular
presentation is used to show the value and limitations
of each method.
INTRODUCTION
This paper compares dynamic kill calculation methods
available to estimate kill requirements during a blowout. Three methods were used in this comparison: 1) _
631

MODEL

The pure friction model uses friction generated by pipe


flow through the geometry of the blowout in the wellbore to establish an equilibrium where the well cannot
produce. When this equilibrium is established the well
is controlled. The combination of pressure losses and
hydrostatic column overcome the resemoir pressure.
The basic assumptions are shown in Figure 1. The
major assumptions are that the wellbore is filled with
kill fluid and the sum of all the pressure losses and
hydrostatic pressures combine to exert a pressure
greater than the static reservoir pressure. The analysis
assumes that pumping operations have been successful in purging the blowout wellbore of reservoir fluids,
so that the kill fluid is not contaminated with reservoir

-rfson

of stem

Sbte W

Tmn~mt

AnaW

WC

Kill Models for

fluids.
With
this
basic
assumption
pressure
calculations can be made with reasonable accuracy
using a variety of pipe flow equations (e.g., power law,
etc.) that are known to match the rheology of the kill
fluid. This approach will yield the minimum kill rates
that fit the assumptions, and allows the kill scenario
to be analyzed without sophisticated
models and
computer capability.
Several significant relief well
projects have been successfully solved using this
approach (e.g., Indonesian blowouts 1980, 1981). The
pumping schedules used employed pumping water
ahead of the kill weighted fluid. Using this approach
the kill rate can be increased or decreased to exert the
desired pressure profile and bottom hole pressure. One
key issue is the possibility of fracturing of the open
hole section. The changeover from a well under control
with water plus friction, to a well under control with
kill weight fluid in the hole can be achieved by altering
the pump rates while displacing with kill weight fluid.
Observations of bottom hole pressures during pumping
are usually a guide to the actual operation whereas
the calculations are used to determine the pumping
spread
and hydraulic
horsepower.
However, the
limitations of using this method are as follows:

Hydraulic horsepower requirements are usually


very high (especially when the well is full of
water only)

Volumes required to kill


known from the analysis

The pressure profiles during the transition


from the well flowing and well under control by
hydrostatic pressure are not known

the

This method will be compared


models in the following section,
STEADY

STATE

TWO-PHASE

well

are

predictionof PumpingRequirements

particular
blowout flowrate
flowchart shown in Figure 2.

is

described

lADC/SPE

in

the

The basic calculations shown in Figure 2 can be


used to determine kill rates required to overcome the
strength of the reservoir, This is accomplished by using
the model in a slightly different way, Kill fluid is mixed
with the produced (wellbore) fluids to determine what
mix rate is required to bring the bottom hole flowing
pressure near to static reservoir conditions, reduce the
amount of reservoir fluid in the wellbore to nil and kill
the well. Mixing ratios of kill fluid to reservoir fluid are
added and a steady state equilibrium is calculated.
This is represented by the flow diagram of Figure 3.
TRANSIENT

TWO-PHASE

FLOW MODEL

The transient two-phase flow model calculates changes


in bottom hole pressure at successive intervals as fluid
is pumped in. Increases in hydrostatic pressure plus
frictional pressure cause a rise in flowing bottom hole
pressure. The IPR is then used to calculate formation
fluid flow volume into the wellbore during the time
period selected. The normal approach is to calculate
the time from the incremental volume and pumping
rate of fluid pumped into the flowing wellbore. This
allows the model to step through a kill using, for example, 20-barrel increments. Once the solution to the
kill is determined at 20-barrel increments, a second
run can be made on the well using 10-barrel increments. If the estimated well response agrees closely for
both cases, there is no need to go to smaller increments. If there is a significant (approximately
10%)
difference in estimated well response, then a fimther
reduction should be made to 5-barrel increments. This
continues until an accurate description of expected well
response is achieved.

not

to more complex

If long flowing times are thought to affect the probability of a successful kill, an actual formation recovery
curve can be built into the model. The worst case
situation is for a formation to recover instantaneously,
and this is assumed in the first calculations. If the
well can be killed in this worst case, then the well can
be killed regardless of the actual recovery time. Should
the model show that a steady state condition is
reached and the well is unloading all the pumped-in
fluid at the same rate it is being injected, then a
reservoir recovery time should be examined to see if the
well kill can be successful.

MODEL

The steady state model accounts for two-phase flow


behavior in a steady state condition. The use of a computer is essential to carry out this iterative method,
since many thousande of calculations are necessary.
This method accounts for compressibility of the twophase fluid under changing conditions of temperature
and pressure. Exit conditions are used to estimate a
pressure in the wellbore. The condition chosen is an
arbitrary value based on experience. Steady state
equations are used to backstep through the input ~
ometry of the blowout. Pressure calculations based on
the exit condition are compared b a reservoir IPR
(Inflow Performance Relationship). If the IPRs FBHP
(Flowing Bottom Hole Pressure) match within a small
tolerance of the calculated pressure profile, a solution
is reached. If there is no match the exit conditions are
adjusted and the calculations repeated. When convergence is obtained the iteration stops and a solution is
reached. The software approach for estimation of a

This program normally plots variables such as tubing pressure, casing pressure, gas flow out, and mud
flow out versus volume pumped. This allows a chart to
be used during pumping to monitor job progress independent of time. In this way, a moderate change in
pump rate from that estimated does not invalidate the
chart as a guide to job performance. Since pumped-in
volume is normally recorded and displayed in real-time
during a job, the monitoring is easier than if some cal-

632

IADCJSPE

culated amount such as elapsed


job progress.

L. W. Abel and D. W. Shackelfod

time is used to judge

adequate results for design purposes while the transient model is a refinement useful during actual field
operations. Both the steady state and the transient
two-phase models yield more information and reduced
horsepower than the pure friction model in dynamic
kill calculations.

The model can be used to look at various injection


rates or to look at different kill fluids or kill fluid
weights. It is also convenient when comparing the effects of different production rates from the well, since
these changes can be plotted as a family of curves prior
to the job. If one of these matches the well response
from the pumping job, then the chances are greater
that the model is accurately describing well response.

COMMENTARY
One basic problem with using computer models is that
assumptions must be made. In a blowout well many
things may be unknown. Exploration wells do not have
test data that can be used to determine reservoir
characteristics. More importantly the flow path of the
blowout is never precisely known. The connective path
to the relief well and its corresponding choking of flow
cannot be known for certain. The leakoff of the kill
fluid to formation is also uncertain. The list goes on
and on. If reasonable estimates of the unknowns are
not made, the results may not match real world conditions. Under-estimation of kill requirements could lead
to disastrous results as there may be only one opportunity to pump to kill. Overkill can waste resources
and cause missed opportunities to kill the well due to
the extreme logistic requirements of bringing extra
pumping horsepower to the wellsite.

COMPARISONS
To compare the three methods outlined above, two
well types have been analyzed using the pure friction,
steady state and transient analysis techniques.
Figure 4 shows a wellbore diagram for Example 1.
This was an actual case where transient analysis was
used to determine the kill requirements. The chart
shown in Figure 5 compares calculated well response
with actual data measured during the pumping job.
The tubing pressure measured is less than that calculated for the first portion of the job, as a friction reducer was used in the water during this period. A decrease in friction of 50% was assumed while the actual
reduction was around 65%. During the job, the actual
choke pressure was dropped more rapidly than the
calculated choke pressure. This is the result of the
formation not recovering pressure instantaneously as
was assumed for the worst case analysis.

Complex models such as the steady and transient


two-phase models can be very sensitive to certain input variables, such as Gas/Oil ratio (GOR), gas composition (e.g., C02 and I-$S content) and many others,
Therefore the model should be tested to determine its
reaction to small changes in certain input parameters.
The authors suggest that many such sensitivity runs
be performed for those input values for which exact
values cannot be determined.

Figure 6 shows the wellbore configuration of a relief


well where the reservoir maximum
deliverability
(Absolute Openhole Flow AOF) was 900 mmscfld. All
three methods of analysis were employed for comparison. Table 1 shows results from each method for two
types of wells: a snub in kill (Figure 4) and a relief well
(Figure 6).

This is a staggering proposition due to the wide


range of possibilities in the field, causing the designer
to examine multiples upon multiples of scenarios. Furthermore, as the complexity of the model increases
more data is needed, including compressibility,
temperature profiles, precise hole geometry, formation fluid
composition, etc. These input values may have serious
impacts on the solution. Complex phenomena such as
two-phase flow, even in the best of controlled circumstances, sometimes do not behave exactly as the model
estimates. In the blowout well nothing is constant and
many parameters change dramatically with time,

Figure 7 is a plot of BHP and Gas Rate versus volume pumped for the second example shown in Figure
6. This shows that the transient analysis wili suggest
lower pumping rates can be used than the steady
state model, but higher volumes of fluid will be
required as the rate is reduced.
Table 1 shows that a pure friction kill requires
higher pump rates as compared to the steady state
and transient models. However, one does not know
this unless all three analyses are performed. Table 1
suggests that there may be little difference between
the steady state and the transient models for the estimated kill requirements for pumping rates and hydraulic horsepower.
However, the transient
model
yields more information for actually running the job in
the field, such as estimated volumes required to kill
the well and the surface pressures while the kill is underway. The conclusion made with the limited study of
this paper is that the steady state model will yield

All this leaves the designer with serious dilemmas


in how to achieve an accurate solution. The only answer to this is experience. The computer clich6 of GIGO
(Garbage In Garbage Out) applies here as well. To
avoid GIGO, the designer must have extensive experience in real world solutions. The experience factor
the engineers ability to make the proper estimates
and assumptions-cannot
be overstated.
This problem of experience is further exacerbated by
the fact that blowouts are very rare and killing them
633

IAIXYSPE

Canpaieon of Steady State arid Transient Analysis Dynamic Kill Models for Preckticm of Pumping Requirements

with a relief well is even more rarely done. This leaves


a limited few with multiple-well experience for dynamic
kill calculations. Fortunately, these experienced engineers are in the consulting and service segments of the
business and their experience can be obtained on a
job-by-job basis. In these life and property-threatening
situations, it is the wisest solution to get experienced
personnel onboard the project team as soon as the
need arises.

PPG

Pounds Per Gallon

P,

Pressure of the reservoir static

PSI

Pressure per square inch

PVT

Pressure Volume Temperature

Q,

Gas flow rate

Pump kill rate

TBG

Tubing

TVD

Tme Vertical

ACKNOWLEDGMENTS

WT

Weight

The authors wishes to thank Halliburton Energy


Services of Duncan, Oklahoma for their cooperation
and participation in this article. The editorial
assistance was essential and the permission to use
HES material is greatly appreciated. Mr. Jerry
Winchester and Don Shackelford of HES Well Control
Division deserve recognition for their valuable
assistance and contributions to this article.

AP

Delta (change) pressure due sum of all hydrostatic, friction, etc.

S1 Units

References
1 Fire#ighting and Blowout Control, Abel, L.W. et al,
ABEL Engineerin~ell
Control Co,, Houston, Texas,
504 pages, 1994.

Nomenclature
Absolute Openhole Flow

Barrels

42 US

BHP

Bottom hole pressure

gallons

BPM

Barrels per minute

CSG

Casing

DC

Drill collar

DP

Drill pipe

FBHP

Flowing Bottom Hole Pressure

ft

Foot or feet

fts

Cubic feet

Gal

Gallons

GIGO

Garbage In Garbage Out

GOR

Gas Oil Ratio

HES

Halliburton Energy Services

HHP

Hydraulic horse power

IPR

Inflow Performance Relationship

in

Inch

Pc

Personal Computer

P,

Pressure combined due to all AP

P,

Pressure due hydrostatic

Bbl

Barrels

X 1.58984

E -01

=m

Gal

gallons

3.78533

E -03

= m

IN

Inch

2.54

E+OO =cm

FT

Foot or feet

3.048

E-01

=m

FT3

Cubic foot

x 2.831685

E -02

= m3

PPG

Pounds per gallon

Pressure per sq. in.

x 6.894757

PSI

2 Dynamic Kill: Controlling Wild Well A New Way,


Blount, Elmo. et al, WORLD Oil, pp 109-126, October
1981.

AOF

Depth

Million standard cubic feet per day

634

1.19829

E +02

= Kg/m3

E +00

= kPa

iADC/SPE

L. W. AIM and D. W. Shackeiford

Pressure

Blowout fbwpath
determirwd

v. Depth

IS

Resewoir parameters
ars deterrnir!.ed (estimated):

(estimated)

IPR, press, temp, PVl data

0
+
%
~

im
~:0
:*

Assutrw exit ccditions


(sonic, or sub-sonic)

s
0

is
:ILl

: w
~;
;*
:*
:C
:<
;m

Set exit conditions to

Program back steps through the geometry


from the ex!t to the reset-vow cakulated

:-!

;=

;0

&
I

Exk velocity
(fbwrate)
is

bdju.stad up or dowr
as needed

::
-.:
:.
:.
,/

.,,

P=

Figure 1
Re$sr-mr pararnstem
Ire dstenr+-dd
SIR, WSM, mmp, PVT datz

S&out fbvmsth k
~
detminsd
(estwnated)

Figure 2

*sumexit c0ndlv0r6
(Sink,

w sub+cmac)

Qk

4
S41 exit ccnditii
to
vtwmrwy V.IW (e.g.
flowrate fixed)

Prcqram bck steps tfwwugh the


g-try
fem the exit !. the
msewdJ estmmii
pmmre PrOfSe

Blowout
well

13-31V CSG

*
I

L
(tokrmm
- Isw.j

L-.

Em vekcny
Ubw7ate) Is
adjustad up or dcwl
aareadsdto
Convares to a
SCMOrl

\
1
13-3/8 @ 2010

DP leak @ 2020

p,
Solution is raacfwd
mmsure am

nd

estmawd
lot
a

Kill rate is
UKreawd, new
exit condtt uns
dctenmmed

?-

4
Kal rate m adJuswd
upwards from hst

7-518* DC

ym

sduE10ni9mach@rbw
u M fm the akulatedM

Q9
Figure 3

Figure 4

635

2-1/16 TBG
Kill String
Gas flow area

lADC/SPE

Comparison of Steady State and Transient Analysis Dynamic Kill Models for Prediction of Pumping Requirements

12

.- \

Gas Flow into Wellbore (relative)


\
%\

10

\
\

~rs

~n

\
\

Calculated Tubing Pressure

I
\

,- -

-# ----

/--c #- o ;HP

Measured Tbg Press


n
norm

-e-

onnoon

. ..--

-----

-%%
-\

-w
\\

~xx-

x .- ~

Calculated Choke Pressure


------ ---------------.-, ___
w MOI>XX2Z %xx.x.x
x,x
-- *-----Measured Choke Press
xJ+=x+gx,x
......x.+__.........+.._.._...+_
. . .
........ .. ....
......... .....+...... .............+... .. ............ 4-.

0
o

50

100

150

200

250

VOLUME

PUMPED

300

350

-\
-------

\
%
-_
N
_+-.~:=\_.

400

- BARRELS

Figure 5

Y
NFLOWPSl~RMCE
REUllONSiPb
au N?Tlmn D

J
W

......

. . . . .
. . . . . .
. . . . . .

,~\

Flwh@osOnl
Iuoh.uwd

Figure 6
636

450

500

lADCLSPE

L. W. Abel and D. W. Shackelford

~RA~~l~w *MALY~l;Note: Rate based on 10,000 psi maximum tubing I essure and is
1400

&L___

120Q

Bottom Hole Pressure


I

om

8C0

g
2
:
a
&

Gas Flow

600

Rate

400

200

200

400

800

600

Volume Pumped InBarrels

Figure 7

Table 1 Kill Comparisons


EXAMPLE 1

TVD = 10,875 ft
Pr = 6700 psi

:
:

BHP
(psi)

! Mud Wt
~
{ppg)

HHP
Ok ; (horse
~ (bpm) ~ pwr)
!

Pure Friction
i
6700
!
8.44
: 21.5U ~ 2087
...................................+ .......................... ..... ...............+................. .....................
12.5
: 3.0 j
Steady State
!
2700
:
530
,.,..,.,,.!,.....!....,......-..+ .......................... ....... .-..
..!..!.:.... .............. ..................
Transient

EXAMPLE 2

TVD = 4920 ft

Pr = 1300 psi

2700 to
6700

BHP
(psi)

\
:

12.5

~ Mud Wt
;
PP9)

:,

3.8-

930

HHP
;Qk:
(horse
~ (bpm) ~ pwr)

Pure Friction
j
1300
;
8.44
!
...................................>.+..
......................... ............ ............

68.7:
2287
...........+.....................

: 16.4 +
680
Steady State
~
992
:
9.2
................4....
................. ........................... ........................... ...............c.y....d...............
Transient

992 to
1300

~
:

9.2

~ 12.9 ~

440

Note: Assumes instantaneous formation pressure recovery BHP = static


Note: Not possible due to high friction loss, list for comparison only
Note: Annulus pressure loss only (e.g. theoretical value only)

637

lWX)

12CCI

Potrebbero piacerti anche