spe-11057-ms
TRANSCRIPT
-
8/18/2019 SPE-11057-MS
1/14
SP
SPE 11 57
Society
of Petroleum Engineers of A M E
Simulation
of the Wellbore
Hydraulics
While
Drilling Including the
Effects
of Fluid
Influxes and
Losses and
Pipe
Washouts
by Keith
K
Millheim
Amoco Production Co.
and Said Sahin Tulga
Drilling Resources
Development Corp.
Members SPE
Copyright 1982 Society of Petroleum Engineers of AIME
This paper was presented at the 57th Annual Fall Technical Conference and Exhibition of the Society of Petroleum Engineers of AIME
held in New Orleans LA Sept. 26-29 1982. The material
IS
subject to correction by the author. Permission to
IS
restricted to
an
abstract of not more than 300 words. Write: 6200
N
Central Expressway P.O. Drawer 64706 Dallas. Texas
This
paper presents a method to simulate
the
circulating system while dr i l l ing a well. For any
given
pump rate
the
pressure losses
through the
sur
face system, down the
dr i l l pipe, through the bi t ,
and up
the
annulus can
be
determined. The algorithm
also has the capabil i ty of simulating a washout in
the dr i l l
str ing,
losing
fluid
to the
formation,
having fluid produced into
the
annulus, and frac
turing
the
formation(s).
The algorithm
is
general
enough to calculate
pressure losses for turbulent and
laminar flow,
simultaneously.
This covers
the
s ituation where
multiple flow
regimes exist
in
the same circulation
loop.
Formulation of the algorithm is
presented,
showing how
a network type of
solut ion
is used to
calculate
the pressures and flows. The i terat ive
solut ion converges rapidly
and can be
used for
real
time
and fas ter than real
time
simulation.
Detailed
surface
pressure
data was obtained
from two wells in
Texas.
The circulation
simulation
program
was
used
to
calculate pressure losses
at
various
depths
in each
well for a variety
of circu
lat ion ra tes . Results
presented in
this paper show
close agreement with
the
fie ld data.
To
show the versat i l i ty of the
simulation
algorithm
a series
of idealized circulation system
simulations are presented.
These
include various
downhole
circulat ion
s ituations
such as
los t
circu
la tion, circulat ion
without
returns, and
fluid pro
duction response
as
a function
of permeabili ty
and
pressure, and
circulat ing
with a hole
in the dr i l l
str ing.
INTRODUCTION
The mud-circulating system
is
one
of
the major
components in
the dril l ing system.
Because of the
paper.
importance of the mud-circulation system, there has
been
widespread
in teres t
in t rying
to
predict
the
pressure
losses
in the system
for
various fluid
types and downhole
wellbore
conditions.
Bobo
,2
and
Moore
3
were
two of the ear l ier
invest igators who developed algorithms
for
the
dril l ing circulation
system
that would
calculate
the
pressure
losses
for a
given
pump ra te . Other inves
t igators l ike Fontenot,4 Zamora,s Denison,S
Schuh,
7
and Kendal
8
presented findings on downhole hydrau
l ics . Some of the
work
concentrated
on a
single
aspect
of
the circulation system, whereas other
studies
referred to the
entire
circulation
system.
Special hydraulics manuals and s lide
rules,9,lO,11 12
and computer
programs were
devel
oped
for
the dr i l l ing person.
Comparisons of
f ield
data with
the
calculated
results
generated from
the various
techniques
did
not always give sat isfactory
results .
In part , this
i s
due
to the oversimplif icat ion
of
the
algorithms
derived
to simulate the mud-circulation system.
Fontenot and
Clark
4
recognized
the need for
including
the
variation in fluid propert ies , well
bore geometries, and dr i l l
str ing
propert ies . They
presented an algorithm designed for the
computer
to
determine the
pressure
losses for a multivariant
mud-circulation
system.
After reviewing the exis ting algorithms i t was
decided that a new approach was necessary to bet ter
simulate the
multivariant
downhole
conditions.
The
basic
idea was
to
develop an algorithm that was
gen
eral
enough
to
simulate almost
any
circulation s i tu-
ation
with
any
type of f luid, wellbore, and
dr i l l
s tring
configuration.
This
paper
is the f i r s t reporting of
the
algor
ithm and
i t s
ut i l izat ion .
CIRCULATION SYSTEM MODELING
The downhole
hydraulic
circulation mechanism is
i l lus t rated
in Fig. l a) .
Drill ing
f luid
pumped
by
the mud pumps travels through the surface equipment
-
8/18/2019 SPE-11057-MS
2/14
2
SIMUL TION
OF THE WELLBORE
HYDR ULICS WHILE DRILLING,
INCLUDING THE EFFECTS OF FLUID
INFLUXES ND LOSSES IN
PIPE
WORKOUTS
1105
and
down
the dr i l l stem to the
bit . Once
the
dr i l l ing
flu id is
out
of
the
bi t i t t ravels through
the annulus between the dri l l string
and wellbore
where
i t
in teracts with
the
geological
formations.
Depending
on the
dri l l ing f luid, formation pore
pressures and
permeabilities,
some
f luid
may
be lost
to
the formations
or some
formation f luid may be
gained in the
wellbore. The
net flow in
the
annulus
is then
diverted into
the
shale shaker
or
through
the
choke manifold. In
addition,
both the blowout
preventer
and choke
manifold
may
be
closed.
I f
this
is done, a casing head pressure
may
develop,
depending
on
the dri l l ing fluid
and
formation pore
pressures.
Another consideration
which affects the
downhole
hydraulics
is the Occurrence and gradual
enlargement of
a washed out
hole
somewhere on
dr i l l
string.
Such
a hole will
divert
some of the flu id
in the dr i l l s t r ing
into
the annulus before i t
reaches the bi t . Another situation is where the
pressure
in the annulus exceeds the formation frac
ture pressures, in which case, some of the forma
t ions
could be fractured
and
substant ial amounts of
dril l ing flu id could
be
lost .
Since the flow rates for the permeable
forma
t ions
and
the
washed
out hole
and subsequently
the
net
flow
in
the
annulus
are
not
known a
pr ior i
a
flu id
network
solution procedure
is needed
to com
pute the pertinent variables.
The
time dependent
solut ion
strategy
is:
1. Obtain
a
steady-state
response
for the
network
model
for
a
particular in terval of
time.
2. Invoke
the
material balance requirement to
calculate the fluid level in
the wellbore
and
the element
fluid properties for that
time in terval .
The
network
model
for the
dr i l l ing
circulation
system which is
made up
of dr i l l stem,
annulus and
geological formation
elements, is shown
in
Fig.
l (b) . In
this model, every dr i l l string pipe,
tool
jo in t col lar
and
annulus
section
is
modeled
as
a
separate
pipe
element with
different
fluid and
geometric propert ies
(see
Fig. 2).
The effective
hydraulic
diameter
for
annulus elements is:
d
a
d - d
h P
1)
The advantage
of
having
pipe
and
annulus ele
ments with different f lu id
proper t ies
is th added
capabil i ty
of being able to simulate the
variat ion
of rheological
properties,
solids
distribution
and
concentration,
and
other
varying
properties. By
using th is approach viscosity sweeps,
pumping
cement
with spacers,
and
other such fluid displacements or
the spotting of f luids can
be simulated.
Also,
influxes of
the
different formation fluids are
han
dled with l i t t l e difficulty.
Each
geological formation with
a
nonzero perme
abi l i ty is modeled
as
a formation
element
with a
pressure loss
character is t ic governed by
the radial
fJow
version of Darcy s law
13
(see
Fjg.
2).
d
3.13 x 10
5
In r Q
kh d
h
A
Q
g
(2)
The formation
fluid
viscosity is
in
function of
temperature and
pressure.
~ T )
= 2.566
-
0.291
T +
1.422 10-
4
T
2
- 3.108 x
10-
7
+
2.4173
x
10-
10
T4
and
for
oil :
~ T , p )
(1 + 0.001
P
P
-9.1228
x
10-
3
T
(5,81153
e
(3)
(4
I t is noted that in the current model only
l iquid
formation flu ids are considered. Incorpora
t ion of
a
gas kick simulation capabil i ty
is
cUr
rently
underway.
The
fundamental relat ionship
used
in
the
anal
ysis of one
dimensional
f luid flow problems is the
Bernoull i s equation modified for the
pressure
los
For a pipe
flow,
Bernoulli s
equation writ ten for
points
1
and
2
is :
14
(see
Fig.
3)
2
In dr i l l ing probl
2
ms, the
magnitude
of kinetic
energy terms
p
(v /2) are negligible as compared to
the other
terms.
Therefore:
p
1
(6
Hence,
the net
pressure
difference between
two
points
in
a
pipe
sect ion
is due to hydrostat ic pre
sure difference and the pressure
loss
between thes
two
points .
In the
current
version of the model,
the
dr i l l ing flu id
is modeled as
a power
law fluid
15
(see
Fig. 4) for
which:
t
(7
The
power
law constant K
and
power factor n can
be
determined graphically
or
calculated from
yield
point
t
and
plast ic viscosity measurements mad
with
a ¥otational viscometer, e ~ i g n e
for
Bingham
plast ic fluids
15
for which:
t
(8
-
8/18/2019 SPE-11057-MS
3/14
11057
KEITH MILLHEIM S. SAHIN TULGA
3
The
values
of K and n
can be calculated
from:
n
3.32
log (9 )
K
100(1022)n
(10)
For
power law
flu ids, the
Reynolds
number and
sure loss for
dr i l l stem
and annulus
flow
l 6
,1 are:
Re
p
Re
a
-n
2a8
b
(n-l)
1-b
Kb [3n + 1J bn
p 4n
Q2-b(2-n)
2-b(2-n)
A Q
p
1-n
p8
2-b(2-n)
Q
A Q2-b(2-n)
a
2-n
Q
(11 )
(12)
(13 )
(14 )
for
which the
Fanning
f r ict ion factor constants a
and
b a r e :
1 6 ,7
24
for
laminar
flow
a
(15)
0.02
log
n 0.0786
for
turbu1ent
flow
for
laminar
flow
b
(16)
0.25
-
0 143
log n for
turbulent
flow
The
cr i t ical Reynolds
numbers
are:
6
Re
cr
3470 - 1370 n
4270
1370 n
for laminar flow
(17)
for turbulent
flow
The pressure
loss
a t the
bi t
can
be
determined from
Bernoull i s equation
using
Equation (18).
(18)
where C
b
is
a
correct ion
factor
used
to incorporate
the f r ict ional pressure loss . A value of 0.95 to
0.98 is
u t i l i zed
for Cb'
For
one-dimensional
f luid
circulation
systems,
there
may
be addit ional pressure losses
due to the
changes in flow
cross sect ion and/or
in flow direc
t ion.
Any change
in
flow
cross
sect ion
or
flow
direct ion dis turbs the
normal
veloci ty dist r ibut ion
and
subsequently
mechanical energy is converted into
heat through the
act ion
of turbulence. Such pres
sure losses
are
cal led minor losses . The name is a
misnomer
however,
because
in
some cases
minor losses
may be more important than f r ict ional losses . The
magnitude
of
minor losses can
be determined ei ther
analytically through the use
of
momentum and Ber
noul l i s equations or experimentally. For the
dr i l l ing circulation system as
explained
in this
paper, the signif icant
minor losses
occur a t the
entrance
and
exi t of tool jOints
and
annulus ele
ments, and a t the
blowout
preventer
and
choke
mani
fold valve res t r i c t ions . Also a
minor pressure
loss
will occur at the bi t exi t due
to
180
0
change
in
flow directi on.
form:
The
minor pressure losses
are
usually in the
2
pM _ Q -
2
gA
(19 )
where M
may
be
a function of
geometry
and/or
flow
parameters.
For
various tool jo in t
secLi
ons, Deni
so n
6
t -
ermined
the
minor
loss coefficient M experimen
ta l ly ,
as
a function of Reynold s number. For the
blowout preventer
and choke
manifold
valve
rest r ict ions M i s analyt ical ly
determined
8
as:
M
J
+
'1
20 A
V
(20)
There are no known
analyt ical
or experimental
results to determine the magnitude of
minor
l o ~ s s
at the
bi t
exi t due
to
the 180
0
change in
flow
direct ion.
Hence
as
an approximation,
experimen
ta l ly determined values
9
for M for pip elbow con
nections are used.
-
8/18/2019 SPE-11057-MS
4/14
-
8/18/2019 SPE-11057-MS
5/14
11057 KEITH MILLHEIM S. SAHIN TULGA
5
where A is a nonsymmetric,
sparse,
square matrix,
whose
elements
are 1, 1 and l inearized pressure
loss
equation slopes
for the
pipe sections.
Q
i s a
vector of unknown
pipe sect ion flow
rates and B is a
vector
storing the
external flow
rates , the hydro
s t a t i c component of pressure
dif ferences
between the
constant pressure points and l inearized pressure
loss
equation
constants for the pipe sections.
Since
nonlinear simultaneous equations
are
to
be
solved
for
the pipe
sect ion
flow
rates ,
an i tera-
t ive solut ion
procedure
must
be used. The solution
technique
ut i l ized
to calculate the
steady state
response
is as follows:
(1) Assume
pipe
section
flow rates Q
(2) Calculate the
pressure loss constants
for
pipe sections,
(3) Linearize
the nonlinear pressure loss
equations using the
assumed
flow
ra tes ,
(4) Calculate the elements of the A matrix and
B
vector,
(5 )
(6)
Solve AlIa
B
for Q
Compare the
calculated Q with
the assumed
Q
for
convergence. I f not converged,
go
to
(2)
with the
newly calculated
Q
as the
assumed
flow rate vector .
I f
converged,
calculate the
pressures.
An Euclidean error norm type of convergence
cr i ter ia is used:
S
< m
(30)
where m is a predetermined convergence constant. In
this
project m 0.01 is
used.
Once the
steady
state response i s calculated,
the material balance requirement i s used to calcu
l a te the f luid
level
and
mater ial
propert ies for
tha t time in terval .
Formation
Fracturing Simulation
After al l the element pressures are calculated,
the
magnitudes
of
annulus
element
pressures are
com
pared
with the corresponding
formation
element frac
ture strengths. The formation element is assumed
fractured
i f :
(31
)
and the formation
element propert ies such
as
permeabili ty,
thickness
e tc . , are revised for
the
next time step,
as determined by
a
separate model.
I f
there
is more than one formation element for
which
Equation
(31)
is
sat i sf ied ,
then only the ele
ment with max(P
FR
-
Pa)
i s modeled as fractured.
I f the annulus element pressure
corresponding
to
an already
f ractured
formation for a par t i cu la r
time step is
less
than
the formation fracture pres
sure, i . e i f
P
a
P
FR
32)
then
the fracture
is
modeled
as closed
and the for
mation
element
assumes
i t s
propert ies
prior
to frac
turing for
the next
time step.
The
assumption
of
incompressible flu id
is no
longer valid for the leak-off
t e s t
simulation
because mud compressibi l i ty
plays
an
important role
in
the
system
response. The dr i l l ing flu id compres
s ib i l i ty
for
a pipe or annulus
element
is given
by:
c
1 tJ V
tJ.p V
(33)
The dr i l l ing flu id
compressibi l i ty
is in general a
function of temperature,
pressure
and
f luid
compo
nent
densi t ies and
volumes.
The solut ion
procedure for the leak-off t e s t
simulation is as
follows:
1. Calculate the casing
head
pressure as:
(34)
where P is the casing
head
pressure and ZQf
are
theCRet
inflow
to the wellbore
al l
calcu
la ted from the previous time
step) .
2. Determine the annulus
pressure
and the
forma
t ion flow rates for the permeable
formations.
3. Check
for
formation
fracturing.
In the
computational
implementation of
the
model, six
different
operat ional
conditions
are
identified
as
cases . This classi f icat ion is
based
on
the status
of
pumps,
blowout preventer,
and
choke
manifold
as
shown
in
Table
1.
Furthermore,
six
subcases
for
each case are identified
based
on
geometrical
considerat ions as to
the
presence and
location
of
permeable formations
and
a
washed out
hole. The
descript ion
of the subcases
and
the order
of
A
matr ices, Band
Q vectors are shown
in Table
2.
COMPARISON
OF FIELD AND CALCULATED RESULTS
Surface
pressure
data from two wells dri l led in
Texas
were
compared with
calculated pump pressures
using the algorithm ci ted in this paper. Table 3
presents the comparison of the field and
calculated
pressure resul ts .
-
8/18/2019 SPE-11057-MS
6/14
SIMULATION
OF THE WELLBORE
HYDRAULICS WHILE
DRILLING,
INCLUDING THE
EFFECTS OF
FLUID
INFLUXES
AND
LOSSES IN PIPE WORKOUTS 110
The pressure
data
was
o b t ~ n e d by
a
dr i l l ing
data logger
at
one-foot
intervals . Various random
depths for each well were
selected
to
calculate
the
pump pressures.
A comparison of
the
measured
pressure
data :is
made with calculated
values using
a standard hydrau
l i cs
computer program
that
is avai lable
via
a
time
sharing
computer service. The
pressure losses
in
the
dr i l l
pipe
and
collars
are
ci ted in
the
f i r s t
column.
The
second
column
presents
the
pressure
losses
through
the bi t ,
and
the third column
is
the
pressure losses in the annulus. The
fourth
column
is a summation
of
the pressure losses. Similar cal
culat ions arc made
using
a standard
hydraulics
cal
culat ing s l ide
rule.
The resul t s from the
hydraulics
algorithm pre
sented
in
this
paper
are
segmented into pressure
losses in the pipe, through the b i t , annular losses ,
and
minor
tool jo in t losses.
The dif ferences in the three sets of resul ts
are summarized in the las t three columns.
The
average percent deviation and the average absolute
percent deviat ion indicate that
the
new algorithm
calculates
pressure losses closer to
the
actual
measured values than the other two methods.
What is interest ing
to note
is some
of the
gen
era] dif ferences in the
calculat ions.
Comparison
of
the pipe and
annular pressure
losses indicates
the
new
algorithm
generally calculates lower
pressure
losses than the other two methods. Except for two
values
in the one
case and
one in the
other,
the
computer
and
sl ide rule calculat ions consistent ly
predict higher surface pressures than observed in
the f ield, whereas the new algorithm has almost an
equal
sp l i l
above
and
below the f ield
measurements.
The
reason for
the closer resul ts is
primari ly
because
the
new algorithm does not make as many sim
plifying assumptions
as the other
hydrauliCS
methods.
Where
flows
arc
laminar
or
turbulent ,
pressure
losses
are
calculated using the appropriate
relat ionships for each
flow regime. Variabi l i ty of
the
dr i l l
st r ing,
col la rs , tool
joints ,
and annulus
are
a l l considered as well as f luid proper t ies.
ote
the
minor pressure losses are small
for most
cases;
however, for
increased
circulat ion rates
and
longer
st r ings,
minor losses could
be
as much
or
more than
the
annulus losses.
USE
OF
THE CIRCULATION ALGORITHM TO snfULATE VAIU OUS
DOWNHOLE SITUATIONS .
The
classic usc of
a
circulat ion algorithm i s
to calculate normal pressure
losses
and ci rcula t ion
rates
for
the dr i l l
st r ing,
bi t ,
and
annulus. In
actual
dr i l l ing
s i tua t ions
formations
are
encoun
tered that have
permeahi l i t ies
that allow
f luids
in
the formation to flow into the annulus or for f luids
in
the annulus
to
flow into
the
formation. Whether
the flow is
an
inf lux or
f luid
production depends on
the pressure
dif ference
between the formation pore
pressure and the
Circulat ing
or s tat ic pressure
opposite
the formation.
Fig.
9 shows a s i tua t ion
where
a
12-1/4
in.
hole
is being dri l led.
A permeable formation is
encountered with
a
pore pressure of
.44 ps i / f t .
Using water as the dr i l l ing
f luid
i t
is
apparent the
formation
f luids
will flow into the annulus unless
the pressure different ial is
reduced
to zero. This
can e achlevecl by two methods for t h ~ s
case:
J) the density
of the
dr i l l ing f luid can be
increased or
(2)
the
pressure loss above
the
form
t ion can
be increased
such that
the
equivalent
ci
culat ing
density
EeD) balances
the
formation
pressure. The
ECD
is
influenced
by
the
mud prope
t ies ,
collar
outside
diameter and hole s ize ,
and
circulat ion
rate .
Fig.
9 shows
for
a given
col la
and
hole size and f luid
proper t ies
how
the
change
circulat ing
rate
for a
given
formation permeabili
can
af fec t
the
inf lux
or
f luid
production of the
dr i l l ing
and
formation f luids. At approximately
300 gpm circulat ion rate
the
ECD
balances the
form
t ion pressure. For ci rcula t ion rates
below
300 gp
the formation produces at
a given
rate , dependent
the
permeability of
the
formation. Above 300 gpm
f luid from the
annulus i s los t into
the formation.
This
example
shows how
the algorithm
can sim
l a te a
si tuat ion that
is
frequently encountered in
the f ield. In
the
real dr i l l ing
case
a f i l t e r ca
could build up and retard the inf lux
or
f luid
pro
duction. This mechanism
i s current ly
being
added
the
algorithm.
In one f ie ld
development
for a
secondary
recovery
project , a
los t c i rcula t ion
problem some
times
doubled
the cost of dr i l l ing the well.
The
simulation of the
problem
is presented by Fig. 10
At approximately 4550 f t a high permeabili ty zone
encounlered
and lhe annular
f luid
is lost into th
zone.
Depending
on
the permeability of the zone,
par t i a l or ful l returns are losl . Fig. 10 shows
percent of f lowline returns as a function of perm
abi l i ty . At a permeahility of 220 md ful l return
are lost (actually the annular f luid
level is 4
f
below the
surface) .
For
250 md
the
f luid level i
at
a depth
of
312
f t . This type of simulation
ma
i t
possible to invest igate
the
si tuat ion where i t
might
be
necessary to dr i l l with a f loal ing mud c
dr i l l ing without returns) . The level
of
the f lu
column
can
be determined for
a given circulat ion
rate,
formation
permeability,
and
f luid
dp llsity.
Using
the
network solut ion
i t would
be possihle
to
simultaneously pump down the dri
I 1
pipe and the
annulus which has to
he done
sometimes to control
the well
from a lower
zone
that could produce.
Fig. shows a par t icular s i tua t ion
where
there
arc
two pcrmeaole intervals
a t
di f ferent po
pressures of .40 ps i / f t and .48 ps i / f t . This
typ
of
si tuat ion is
one of the most d i f f i cu l t
dr i l l in
prohlems to encounter. Dril l ing
with
a f luid
den
s i ty
of water
hath
A and B zones
can
produce.
Increasing the mud weight to 9.0 ppg causes zone
to
lose circulat ion
and
zone
B to
s t i l l
produce.
The amount of influx and production is a function
flow
capacity
(kh). Tf
the
flow capaci ty is such
tha t
the
influx into
interval A was 100
gpm
(50 m
i t
might
oe
possible
to
mix
enough
weighted
mud t
conLJnue
on wi
Lhout
prematurely
set t ing casing.
Conversely
the permeability
could
be
too high
whe
i t
would
be impossible
to
continue dr i l l ing . Cas
would
have to be run. The
main
point i s th is typ
of simulation algorithm makes t possible to anal
complex
dr i l l ing si tuat ions that are handled hy
experience and t r i a l and error . Again
i t
should
remembered that the algorithm does not include th
effects of
f i l l e r
cake buildup which would genera
a skin and could reduce the f luid influx
or f l l l i
production.
-
8/18/2019 SPE-11057-MS
7/14
11057
KEITH
MILLHEIM S. S RIN TULG
7
I t
can
be argued that permeabil i t ies and
other
information for doing this type
of analysis
i s di f-
f icu l t to
obtain. However, the
analysis
of good
dr i l l ing
data from
a dr i l l ing data logger) coupled
with logs and other information make
i t possible
to
bracket
the
unknown
formation
properties
in most
cases where
a
reasonable simulation study can be
made.
This algorithm
also
can
be used to simulate
mult if luid circulat ion such as used for cementing,
well control, viscosi ty sweeps, and other such
cases.
Another case that
can be
simulated
i s
when ci r -
culat ion
is
stopped the formations
ei ther
take
f luids
or
produce.
If
the annulus is
par t ia l ly
rest r icted or closed, the algorithm
can also simu
l a te that si tuation. Although not
act ive
yet , the
algorithm
can
simulate
reverse circulation.
Almost every
possible
downhole
circulat ion
si tuat ion
can be simulated using the network type
of
solut ion.
Future work
will include the effects
of
f luid f i l t r a te loss
and
f i l t e r cake
buildup,
t ran
sient
fluid
flow
for
both
radial
and
l inear
flow
geometries, and the
handling
of the more complex
f luid types.
The causality diagram
for
the
mud
circulation
system simulation module is presented by Fig.
12.
SUMM RY
ND
CONCLUSIONS
In th is
report ,
a technique
to
analyze and
sim
ulate
the complete downhole hydraulics
of
a dr i l l ing
operation
is
presented. This
technique
is
based
on
f luid network analysis methods
and is capable of
modeling the
interact ion
of the dr i l l ing mud with
geological formation flu ids, formation fractures and
washed
out
holes which
may
form
in the
dri l l
s t r ing .
Dril l ing mud
is
modeled
as a power law
f luid.
Both
f r ict ional and
minor pressure losses are
included in the
analysis .
I t
is
observed that
1) pressure losses are
not very
sensi t ive to
small
changes
in K and
n; i . e . for small changes in
K and
n,
changes
in pressure
loss
are small, and
2) minor
pressure losses
are
important and should be included
in
pressure loss calculations.
All
configurational
possib i l i t ies such as
blowout preventers and the choke manifold being open
and/or closed are modeled using the new technique.
Only
l iquid
formation
f luids are
considered
in
the present
analysis .
Modeling gas production and
a i r
dr i l l ing
are logical extensions of the
technique
described in th is
report .
Sample
case
runs as compared to actual field
results show that the new
technique
predicts
the
actual pressure losses
closer
than the
other
hydraulic calculation methods
and
will be
of
value
in
the
f ie ld especial ly
for dr i l l ing in
problem for
mations. Moreover, the algorithm will le t the
dr i l l ing person study the effects of the control
lable hydraulic parameters,
such as
bi t
j e t
size,
pump
flow
rate, mud density and viscosity and devise
the
most desirable
mud circulation program. Also,
the
algorithm
forms a
basis for the
study and
formu
la t ion of
new
methods to aid the engineer
such as
in
mUltiple
fracturing, gas
kick
control ,
and the det
ermination of
a
washed out
hole
location.
NOHENCL TURE
a
Fanning f r i c t ion factor constant
A
Coeff icient
matrix
v
b
B
c
C
Cb
D
d
a
d
r
d
w
e
f
g
L
a
L
P
k
K
M
n
N
Total
bi t
j e t
area,
sq. f t
Choke manifold pipe
area,
sq. f t
Choke manifold valve opening
area,
sq.
f t
Fanning f r i c t ion factor constant
Vector containing
external
flow rates
l inear-
ized
pressure loss
constants and hydrostatic
pressure
differences
Number of constant
pressure points
-1
Drill ing fluid
compressibi l i ty,
psf
Bit pressure loss correct ion
factor
Depth, t
Annulus element effective hydraulic diameter,
t
Annulus
element
outer
diameter,
f t
Annulus
element
outer
diameter,
f t
Pipe element
inside
diameter,
f t
Pipe element outside
diameter,
f t
Radial formation bed outer diameter, f t
Washout hole
hydraulic
diameter,
f t
Linearized
pressure
loss equation slope
Linearized
pressure loss
equation constant
Gravi tat ional
accelerat ion
= 32.17
f t /sec
2
Geological formation element thickness,
f t
Pipe element length, f t
Counter for pipe,
annulus and formation
flow
elements
Number
of junct ion points
Geological formation element
permeabili ty, md
Fluid
consti tu tive law constant
Number
of
minimum
loops
Minor
loss
coefficient
Fluid consti tu tive Jaw
factor
order of A matrix
Number of
permeable
formations
-
8/18/2019 SPE-11057-MS
8/14
8
SIMULATION OF
THE WELLBORE
HYDRAULICS
WHILE DRILLING,
INCLUDING
THE
EFFECTS OF
FLUID INFLUXES AND
LOSSES IN
PIPE
WORKOUTS
110
p
p.
ressure, ps
Annulus element pressure referred to the
depth of the annulus element, psf
Casing head
pressure
Casing head
pressure
for previous
i terat ion
Geological formation element fracture pres
sure,
psf
Formation pressure
Formation element wellbore
pressure,
psf
Frictional pressure loss
at
b i t psf
Frict ional
pressure loss at
the annulus ele
ments, psf
=
Minor pressure
loss,
psf
=
Fric t ional pressure
loss at
dr i l l
stem ele
ments, psf
Q
Flow
rate,
cfs
Re
cr
Re
p
=
Flow
rate
vector, cfs
Calculated flow
rates
from previous i te ra -
t ion,
cfs
Pump
flow
rate,
cfs
Washout hole
flow
rate,
cfs
Reynolds number
for
annulus element
Crit ical Reynolds number
Reynolds
number
for
dri l l
stem
element
r Number
of
pipes
S
Ratio of flow rate
Euclidean
vector errOr
norm to current
flow
rate
Euclidean
vector
norm
t Time,
sec
T
Temperature, of
v
Velocity,
ft /sec
V
Volume, f t
3
Washout hole diameter time
constant coeffi
cient
Washout
hole diameter time
constant coeff i
cient
Y
Fluid
shear s t ra in rate
a
=Annulus element
pressure loss
constant
A Dril l
stem element pressure loss constant
p
oss
p
Fluid
density, pcf
Tt Phi number
T
Fluid shear s t ress psf
T
Y
Fluid yield
point,
psf
r Time
constant for
washed out hole diameter
enlargement
IJ
f
Geological
formation fluid viscosity as
fun
t ion
of
temperature and
pressure,
cp
Drill ing fluid plast ic
viscosity ,
cp
The authors would
l ike to thank
Amoco Produc
t ion Company
for
the permission to prepare and wri
this
paper.
1.
2.
3.
4.
Bobo, R.
A.
Drill ing Cheaper with Lower
Pum
Pressures,
Sept.
11
1967,
pp.
Bobo,
R. A.
Current
Practices in Drill ing
Hydraulics,
Jan-Feb
1969, pp.
Moore, P.
L., Five
Factors
That
Affect
Drill ing Rate, The Oil and Gas
Oct. 6,
1958.
Fontenot,
J.
E.
and
Clark,
R.
K.
An Improve
Method for Calculating
Swab
and
Surge Pressur
and Circulating Pressures in a Drill ing
Well,
of
Petroleum
Journal
5 . Zamora, M. and Lord, D.
1 .
Practical
Analys
of
Drill ing Mud Flow
in
Pipes and
Annuli, pr
sented at the 49th Annual
Fall
Meeting of the
Society
of
Petroleum
Engineers
of AIME Hou-
ston, Texas, Oct. 6-9, 1974.
6.
Denison,
E. B., Pressure Losses Inside
Tool
Joints ,
1
Sept. 26,
1977,
pp.
7. Schuh, F. J . Computer Makes Surge-Pressure
Calculations Useful,
The Oil
and
Gas Journal
Aug. 3,
1964,
pp.
96-104.
8. Kendall, H.
A.
and Goins, W C., J r . Design
and Operation of Jet-Bit Programs for Maximum
Hydraulic
, Impact Force or
Jet
Velocity,
Transactions
AIME
Vol. 219,
9.
Hughes Tool Co
10. Hydraulics
for
Jet Bits, Hughes Tool Co. (Jan
1956).
-
8/18/2019 SPE-11057-MS
9/14
11057
KEITH
MILLHEIM S. SAHIN TULGA
11. Reed Roller
Bit
Co., Hou-
12. Hydraulic Calculator, Security
Engineering
Division,
Dallas,
Texas.
13. Dake, L. P.,
Fundamentals
of Reservoir Engi
neering, Elsevier,
1978.
14. Streeter,
V
L. and Wylie, E. B., Fluid
Mechanics, Seventh
Edition,
McGraw-Hill,
1979.
15
Wilkinson W L., Non-Newtonian
Fluids,
Per
gamon Press, 1960.
1 Metzner,
A
B. and Reed, J. C., Flow
of
Non
Newtonian Fluids,
Correlation of the Laminar,
Transition and Turbulent Flow Regimes,
Journal,
December, 1955, p. 434.
17.
Savins, J. G., Generalized Newtonian Pseudo
plast ic Flow
in Stationary
Pipes
and
Annulus,
_______________
______ ____
, 1958,
p.
325.
18. Vennard, J K. Elementary
Fluid
Mechanics,
John Wiley
Sons,
1970.
19. Pressure
Losses
in
Smooth
Pipe
9
-
8/18/2019 SPE-11057-MS
10/14
fo":1l111Wt11 I
~ I ' lflt. l J
hP
CASE
PUMP
NUMBER
STATUS
1
ON
2
OFF
3
ON
4
OFF
5
OFF
6
ON
SUBCASE
PERMEABLE
NUMBER
FORMATIONS
1
2
3
4
5
6
flOW'i
NO
NO
YES
YES
YES
YES
Rf.SUUs FkoH
txl l1lNG
~ ~ ~ ~ ~ ~ ~ ~ ~ ~ I
I I 1 ~
DEp'r"
JUS
RATE
i t,)
l in ' .
11\.) (gplII) Al'pIP:'
¢ P I l : ~ ~
1 2 ~
Il
,00
17.25
"
)-12-11 ,0 )
45'S 1161
"-12-13 l)2
S
2 ~ 7 0
"
-IJ - lJ
J
839
15el
69
i ) - l J - l J
,24
900
1469
.
O· l l ·U
125
'"
1554
"
- 13
I
' \1
'"
.
IJ
.
115
1408
"
·
1 5 ~ I S
24.
'
.
"
-15-15
'IS
'0.
IOU
"
o I l ~
14
12Z
1000
131)0
136
j.ltllll(l
-(AI
t
) .. 1,,1, , , ,1
ho .
S U ~ q ~ ~ ~ . t · l I t Ilydraul rll 'Tab II'S
2260
12:i6
2Z87
2489
2 4 6 ~
20
.
2181
• 09
2101
2'.36
TABLE 1
BLOWOUT PREVENTER
STATUS
OPEN
CLOSED
CLOSED
CLOSED
OPEN
CLOSED
TABLE
2
LOCATION OF
PERMEABLE
FORMATIONS
-
ANYWHERE
BELOW
THE
WASHOUT HOLE
ANYWHERE. ONE
PERMEABLE
FORMATION
IS COINCIDENT
WITH
THE WASHOUT HOLE
ANYWHERE. NO
PERMEABLE FORMATION
IS
COINCIOENT WITH THE
'
01
0
111
190
.
'
.
,
895
8"
WASI«lUT
HOLE
TABLE
3
Rt;SIII,TS FMH
HVDRJ\UI.ICS
______
Illi ;
1132
1168
1616
1574
1"68
1570
465
IoiIlS
1 28
1044
1)03
"
"
.
.,
101
53
33
8.
50
II'
I
l 'v
nn
2282
24)8
V59
2081
,,.
2162'
'"
O S 4
2216
WASHOUT
HOLE
NO
YES
NO
YES
YES
YES
'"
2
299
'"
62
m
11>
5)8
m
1
9"'9
1116
1113
1(;11$
)srs
1412
I ~ 1 ~
46'
11t 2
.
041
CHOKE MANIFOLD
STATUS
CLOSED
CLOSED
OPEN
OPEN
CLOSED
CLOSED
ORDER OF SYSTEM
MATRICES SKETCH
1
4
2*N
f
+l
~
2*N
f
+4
~
•
2*N
f
+3
~
2*N
f
+4
~
X,UEVIATJtJr-
E,JIiI;tin8
.
fool
Ju luh
Hyduul/otl;
Hydraulilt-:
24
I'
.
56
1
J.
2.
'0
J8
.
AI8Qrilh."" TabJ .
20S9 I),
20';11
102
20S) 14 2160 4.4
2001
It
20}1
\0.4
2281 28 2010
23.8
BlS
]0
2(151
202
I Hl 12 :1'0111 - o . ~
669 8
i9) -4.1
2010 22 2081
' . 8
840
11
;8'1 152
186l )4 1816 1:Z.J
2 81 4
2080
17.1
~ v e r a s l l n
-
8/18/2019 SPE-11057-MS
11/14
TO SHALE SHAKER
......
.
BLOWOUT
PREVENTORS
Fig.
1a-Diagram of
a typical drilling circulation loop.
KNOWN
PR SSUR
FORMATION
POR
PRESSURES
Known
at all times)
Fig. 1
b-Nodal
diagram of a drilling circulation
loop.
2
Fig.
3-Bernoulli s
equation written for points 1 and 2
at
depths
01
and
D2.
1
y
Fig.
4-Drilling
fluid models.
a)
e)
T
I
b)
Fig.
2-Element
descriptions.
BINGH M
PL STIC
FLUID
_
POWER L W FLUID
NEWTONI N FLUID
-
8/18/2019 SPE-11057-MS
12/14
0.1 d
I
r
Fig.
5· Washed out hole diameter
as
a function of time.
Fig.
6a Washed
out hole case.
~
~ C O N S T N T
PRESSURE
?POINTS
JUNCTION
POINTS
f
I
Fig. 7 Definition
of flow network.
t
1.0 . - - - - r - - ~ - - - r - - - - - _ - _ = - - - - .
..
0.8
P
3
PI 0.6
. ~ M
0
1
2
e ;;:z
1
pM
0.4
0.2
o
o
0.2 0.4 0.6
0.8 1.0
02
1
Fig.
6b Washed
out hole minor loss approximation.
Q
A 2 -b
2
-n)
gA + Q
QO
P
Q
Fig. 8 Linearization of Eqs. 5 and 26.
-
8/18/2019 SPE-11057-MS
13/14
:E
c..
(, )
x
:=l
.....I
u..
z
0::::
0
:s:
9
u..
:z
0
i=
<
:E
0::::
E2
V'l
:z
0::
::::J
t ;
0::
I.I.J
z
....J
3:
9
u..
u..
0
f5
u
0::
I.I.J
Q..
+1.5
+1.0
+
.5
0
-.5
-1.0
0
100
80
Q
8"
COLLARS
6
-\ (44
psi/ft)
.,
. '... .
12.25 r-
~
........
''=.
. .
'
.
' . ,
. ,
.
50MD
lOOMD
150MD
200MD
100
200
300
400
500
PUMP
RATE Q -
GPM
Fig. 9-Example of fluid influx and production as a function of the ECD.
4F t@ /
0
PUMP RATE AT SURFACE '956 gpm
MUD PROPERTl
ES
:
220
md
316 Ft
@
YP •
61b/100
250 md
1
V • 8 cp
DENS
ITY •
9.0
ppg
JET
SIZES: 14-14-14
2
0
f T 1
0
t
::I:
0
.
);>
Z
Z
c:::
r-
60
»
\
PUMP PRESSURE (psi)
:::0
4-lItORlll
~ 4 2 5
PIPE
.
r
c:::
~ 4 2 5 1 4 2 3
0
40
.....
3
r-
',-1422
f T 1
<
f T 1
1421
r-
' .
......
-,
§
20
"-
4
1420
.
r+
,
~ 8
0
0
100
200
300
PERMEABILITY (MO) OF
LOST
CIRCULATION ZONE
Fig. 10-Example of lost circulation
as
a function of the permeability of the lost circulation zone.
-
8/18/2019 SPE-11057-MS
14/14
I
GEOLOGY
DATABASE
I
RILLSTEM
DATABASE
Q 300 gpm
.- -
4-112 DRILL PIPE
LEGEND
FORMATION
A
FORMATION
B
300
E
200
Q
E
s:
100
u..
-100
-200
-300
o
~ r .
9000 tt
8 COUARS< x ---
8.3
ppg
8.3 ppg
9.0 ppg
_ _
__
x--
>l
--
9.5
ppg
9 ppg
50 100 150
200
250
300
PERMEAB
I
L1TY MO)
Fig. 11-Example of circulation with two permeable formations at different pore
pressures.
I
MUD
PUMPS
PUMP FLOW
RATE
.
I FORMATION
PROPERTIES
OPERATOR
CREATION
OF
WASHOUT
HOLE
I
r : ~ E T R A
TION
RATE
IT FLOW RATE
PUMP
PRESSURE
DEPTH
I:
T
PRESSURE
LOSS
BOTTOM HOLE
PRESSURE
DOWNHOLE
HYDRAULICS
FLAGS
FOR
BOP
AND
eM
STATUS
MODEL
I ·
RILLS'l'EM
PROPERTIES
FLUID I lFORMATION
PROPERTIES FLOW
RATES
EI L BALANCE
Fig. 12-Causality diagram of the hydraulics algorithm
in
the total drilling simulator.