Problem During Tripping
Problem During Tripping
Problem During Tripping
,tiiiMis R
SPE 26330
This paper WM aalacfad for preaenlatlw by an SPE PrQra!n CommltIee following review 01 Info!matlon mntslned [n an sbafracl eubmftted by It!a author(a). Cements of the pupor,
cc preaantad, have not baan ratiowod by the Soclafy of Petroleum Engineara and are eubject to correction by the autfIor(8). The material, ea pre$emad, Ooea not neceeearlly reflect
any poaitlon of tha SOClolyof Patrolmim Englneara, Ha offlcara, or mambera. Papare pra$enled at SPE maatinge are subject to publlcallon mvlow by Edllonal Committee 01 the Society
al Patrolaum Er@iaam. Pormlaalon 10copy la faatrfctad to an abatracf of not more than 300 words. Illuatmtlona may not be copied. The abstract ehoutd contain canaplcuoue ocknowfedgman!
of whar8 #nd by whom Iho paper la preeanted. Write Ubrarian, SPE, P.O. Sox &M3s, Richardson, TX 7MN3.SS3S, U.S.A. Telex, 1S2246 SPEUT.
163
●
2 Problem Detection During Tkipping Operations In Horizontal and Dkectional Wells SPE 26330
164
SPE 26330 Jos4 Venhcio Cardoso Jr.; Eric E. Maidla; Luiz S, Idagawa 3
analysis only the CTA (Center Third Av- As At is the only variable in equation 2, the initial
erage) needs to be calculated and stored in upward trend in figure 10 is therefore explained.
memory. As soon as the increment in axial force ex-
ceeds the differential sticking drag force, the pipe
2) Ledge Type Curve Standard - LTCS In
is released and a normal pattern takes place.
the previous type curve description a question
This occurs because differential sticking does
remained unanswered: Why is a 3 Hz sampling
not happen instantaneously, but rather is a time
rate necessary, if CTA is relatively independent
dependent static phenomena that takes place be-
of this, and acceleration and deceleration effects
tween the pipe and mud cake surfaces (due to the
are known to take place only at the beginning and
cake thickness). This provides a contact area that
ending of a pulled out section?
does not see, any longer, the wellbore pressure.
Notice in figure 8, that there are two distur-
The pressure at the interface does not change im-
bances within the more central part of the type
mediately but rather shifts from wellbore pressure
curve - one at about 3070 of the section trip time,
to the pressure at the rock/cake interface, as a
and the other at about 70?40.In this case it is sim-
function of time.
ulating the tool joint hitting against the ledge,
At the initial stage. of the problem (low pres-
but it could be any part of the drillstring with
sure differential and/or snail times in which the
shoulders on it (e.g., stabilizers). This can only
pipe remains static, and in contact with the cake)
be clearly detected if high sample rates are used.
it is very hard to detect without using high sam-
We found 3 Hz to be adequate.
pling rates, due to the small time in which its ef-
3) Borehole Closure Type Curve Standard fect on hook load takes place (only a few seconds
- BCTCS This problem is well known in the to begin with).
field, and a typical BCTCS is shown in figure 9. The detection of this problem is easier as t!ie
The operation is followed by a succession of mud weight program is depth dependent. As the
pipe stretching, quick sudden movement (high ac- differential pressures are related to depth, the
celeration and deceleration), and further stretch- problem becomes worse as the well gets deeper.
ing, throughout the entire section being pulled This provides enough data to identify this prob-
out, thus indicating much difficulty in all in- lem during the trip operations that normally oc-
stances in moving pipe, Notice that the entire cur. In this case, as trips are recorded at different
string is moving, otherwise the hook load would times, but the different type curves at the same
simply increase as pipe is stretched, and no de- depth, could be used to identify how severe is the
crease would be observed, problem really becoming.
The assumption here is that the pipe contin-
5) Incorrect Operating Procedure Type
uously moves upwards.
Curve Standard - IOPTCS In this case
4) Differential Sticking Type Curve Stan- there are several accelerations and decelerations
dard - DSTCS Initially the pipe only (figure 11), meaning that the drillstring is pulled
stretches according to: out in a “bumpy” fashion. In not using the type
AFH At? ~ ~ curve technique, or not observing the operation,
u =tx E==$— — =G (1)
A. could otherwise lead to the wrong conclusion -
that there might be a problem!
assuming the pipe is picked up at a constant v~ This is detected by the use of type curve anal-
Iocity and stretched: ysis because the hook load lower values exceed
E predicted ones for normal fb values (.f~.). Also a
AF1t = —xv. xAt (2)
A, X !dp larger amplitude occurs, even for wells with prob-
lems.
Ae
where: VH = — (3)
At
4 Problem Detection During “IMpping Operations In Horizontal and Directional Wells SPE 26330
6) Further Analysis Type Curve Standard – to some degree in nearly all measurements, even
FA’JWS ‘Ilk graph is similar to the one shown when they are so called “carefully recorded”.
in figure 7 with the exception that CTA is greater Notice that for this simple example problem,
than the hook load predictions when using the the DWR is calculated as:
normal borehole friction factor (between 0.20 and .
DWR = fbn x tan a not val]d for w& (6)
0.4).
Note: the next step is to be carried out only therefore, for this example, the inclination angle
for directional wells for which the DWR is greater that corresponds to a DWR of 15% is roughly 23.2
than 15% (explain&d later). degrees. In recalculating the pseudo borehole fric-
tion factor yields a value of 0.42, or an error in its
Pseudo Friction Factor Analysis: estimation of 3970 - a significant error decrease,
but still too large at times. A method for cali-
This part of the analysis isn’t intended for
bration checks is presented later in this paper to
the vertical wells or depth intervals of directional
mitigate this effect.
wells for which the DWR is lower than 15!Z0.
Important Notice: The above cannot be ap-
The Drag Weight Ratio (DWR) was first de-
plied to directional wells. There is a consider-
fined by Maidla8, and was slightly changed here.
able vertical portion before the kick off point and
It is defined as the ratio between drag (calculated
inclination changes with depth. Therefore DWR
for a normal borehole friction factor value*) and
should be calculated with a computer program for
the buoyant weight of the pipe:
the entire well profile, and only for those depths
for which DWR > 15% should the pseudo fric-
tion factor analysis be performed (DWR is not
linearly proportional to inclination - it depends,
Therefore if drag isn’t a significant portion of your among other factors, on the well trajectory, as
hook load, and therefore of your buoyant pipe considered by equation 4), unless calibration is to
weight, other small effects on hook load will offset be checked.
the values of& calculated without the well really
having a problem. Just as an example, consider a Analysis Procedure
straight pipe resting against one inclined plane - 1. For every depth below the DWR limit depth,
figure 13. The inclination is 10 deg, the hook load and for each section of pipe pulled out, a
error is 570, and a borehole friction factor of 0.35 pseudo friction factor is calculated or as-
is used. The pseudo friction factor calculated for signed according to the following rule:
this situation is: ● If the hook load analysis indicates the
absence of problems fb= .fb. k as-
fbp = fbn+ & + fb. x I?HL (5)
signed.
= 0.35+ - + 0.35x 0.05 ● If the hook ioad analysis calls for the
pseudo friction factor analysis, then fbp
fbp = 0.35+ 0.28 i- 0.02 = 0.65 is calculated using the CTA value.
This error (89%) becomes a problem if it is sys- 2. A log plot is produced: fbpvs. ~.
tematic, because it would indicate a problem that
3. Pseudo friction type curve matching is used
really does not exist. If the error is of random
tO verify if the vaiues of fbp follow any pat-
nature, the average values would remain around
tern (Idagawag) simulated for:
0.35. Unfortunately in the field, both errors are
taking place, The systematic error is present ● partial borehole closure.
● bridge.
s 0.35 is used here. ● key seat.
166
.
SPE 26330 JOS4 Venikmio Cardoso Jr.; Eric E. Maidla; Luiz S. Idagawa 5
4. The best fit indicates the probable prob!em. The phe,lomena is modeled as:
!5. If no fit occurs? the followirig might explain: (~5&P
AFa = x t~c x KBC (7)
(a) the ‘signatures” (explained later) or .Dbi~
trends are the results of two or more F. = (FJO + AFa (8)
effects. where: (F.)O = Pipe axial load at (DBc)tOPwith-
(b) systematic errors in hook load readings. out considering the partial borehole closure effect.
The type of sensor used and the way For this study, a sensitivity analysis that took
it is calibrated are very important for into account the magnitude of the values for the
this analysis - refer to Maidla’ss paper field data recorded suggested this coefficient to be
for 3 appropriate sensors, for hook load 20 lbf/ft (292 N/m).
measurements, that can be used (there The last item missing in equation 8 to run
are others, but are not the scope of this the type curve simulator is l~c (explained soon).
paper). For now, we will work with a sketch as shown in
(c) a problem occurred that was not mod- figure 14.
eled. Analyzing the sketch it is seen that as the drill
collars, or stabilizers, or any other large diameter
Pseudo Friction
Factor Type Curve for Par- portion of the drillstring (even tool joints - not
tial Borehole Closure A partial borehole clo- actually discussed here since the problem would
sure is illustrated in figure 2. A long section already be very severe) starts passing through the
of pipe suffers the action of a lateral force, dis- partially closed section, there will be a sudden in-
tributed in some manner over its length, that in crease (could be small) in hook load (point D)
turn increases the hook load. that will gradually increase to an average value
The following assumptions were made for (point C) that will remain fairly constant un-
modeling of the phenomena that is further refined til the bit reaches the bottom of the partially
to a type curve display: closed section (point B) and after decreasing in
The hook load increase is directly propor- an unknown manner as the bit starts passing
tional to the length of the borehole closure through, until finally reaching the partial bore-
SeCtiOn(ff3c). hole closure top (point A) and .fbp returns to nOr-
The effect takes place as long as there is pipe mal (point O).
below or within this section. In this paper, a term signature is used to de-
scribe the behavior of the borehole friction factor
It’s effect is time dependent in some way -
log. For this case, the probable problem signature
the longer it takes to get by it, the greater
is shown by the lines connecting OABCD13.
the effect. This is modeled by assuming
At this point, item 4 of the “analysis proce-
that the additional increase in hook load will
dure”, can be further detailed:
be proportional to the initial depth of the
partial borehole closure and the current bit Once the meudo friction factor log is ready,
depth. figure 14 ;S used qualitatively to ;heck and
see if there is any similarity between them.
An empirical coefficient (KBC) will provide
the ne&ssary degree of’ accuracy required ● If the possibility exists, points A and B will
to account for many effects that are not be tentatively defined on the log, and this
considered independently (since knowledge wi)l define the borehole closure length (.f.Ec).
of exact geometry is not possible) such as: This is the last value missing to run the type
surface roughness, borehole friction factor curve simulator.
within the partially closed area, contact ● By running it, a scaled type curve is pro-
forces due to the geometry changes, etc. duced and provides the best analysis tool.
167
6 Problem Detection During Tripping Operations In Horizontal and Directional Wells SPE 26330
Sensitivity analysis is then performed to try check this, the value of DB can be tentatively cho-
and verify if both signatures match to some sen and an adjustment of the ]{B value should be
degree. tried in order to obtain the best match possible,
interactively - this is done quickly with the sim-
Pseudo Friction Factor Type Curve for
ulator.
Bridging A bridge is illustrated in figure 6. A
short section of pipe suffers the action of a lateral Pseudo Friction Factor Type Curve for Key
force, Seat A key seat is illustrated in figure 4, Part
The following assumptions were made for of the borehole is worn out by some component
modeling purposes: of the drillstring (e.g., the drillpipe).
● At the moment the bridge occurs, the hook The following assumptions are made for mod-
load would suddenly increase by some value eling the hook load response:
(in a step wise way), here modeled as: The effect decreases as the drillstring is
DEi pulled out. The reason being that for this
= z~Bx
(A~’)initia( ~Dbit)initial (9) situation the normal force between pipe and
the key seat surface is decreasing as pipe is
The effect takes place as long as there is pipe
● pulled out.
below the bridge. Notice: here it is not as- The key seat effect is less severe for larger
sumed that the bridge effect will necessar- radius around which the drillstring is be-
ily cease as the bit reaches the bridge, this ing pulled, The reasoning in this case is
could happen at any moment depending on that doglegs generate larger normal forces,
the severity of the problem. as shown by any drag model* ‘2’3’6.
● The effect is time dependent in some way -
The phenomena is modeled as:
the longer the time ii exists for, the greater
D
the effect. This is modeled by assuming that
the additional increase in hook load will be
F. = (F. ). x e ( b“‘D k‘x]’’”) (11)
168
.
SPE 26330 JOS4 Ven&ncio Cardoso Jr.; Eric E. Maidlx Luiz S. Idagawa 7
key seat problem will be within points E and C. WP,(/) = principal component of distributed
For depths above point E, the & might not have weight, lbf/ft.
even been calculated (if the suggestion for calcu-
lations of& previously described is followed). and
Assuming some variation in the pulling veloc- creases. This is the first indication of the possibil-
ity (it is lower toward the end) this plot is sim- ity of a systematic error - if this decrease doesn’t
ilar to the ledge type curve standard shown in occur (for an average fbp value above 0.4), the cal-
figure 8. ibration checks can’t be performed since it means
To confirm this hypothesis, this behavior that the effect of systematic errors is increasing,
should repeat itself throughout the other sections that isn’t reasonable (if all equipment is in good
pulled out - this was not observed. Therefore working order), and therefore some kind of prob-
this was not diagnosed as a drilling problem (this lem might be occurring in the well.
is not the same as saying that no ledges existed 2) If on the other hand, the average fbp value
in the well). is below 0,25 and the log values tended to increase
Further seen is that the pseudo friction factor to values close to 0.3 or 0.4 [not more than this),
values, that would be calculated using each hook this would also indicate the possibility of some
load point in figure 20, has an average close to systematic error in the measurements.
0.3, that is a normal value expected.
CONCLUSIONS
2) Pseudo Friction Factor Analysis:
The field case study supports the use of the
A typical hook load plot, recorded at 3 Hz is Type Curve )kf@hing Technique, that includes
shown in figure 21. The acceleration affect at the
the “Hook Load Data Analysis” and the “Pseudo
beginning and deceleration at the end are clearly
Friction Factor Analysis” (used here as a calibra-
noticeable, as is the quality of the data in the
tion checkup tool), for the purpose of Borehole
center third average of the plot.
Diagnosis).
The hook load type curve analysis showed no
The results showed that the graphical ap-
need for type curve pseudo friction factor analy-
proach is useful for the diagnostic process,
sis as the DWRS were below 15Y0. The analysis
whether to build type curves or to analyze pseudo
performed was for calibration checking.
friction factor logs. It showed to be equally useful
Systematic measuring errors cause large errors
while checking equipment measurements calibra-
in & if DWR < 15~o,making this a good tool to
tion while performing sensitivity analysis.
check out calibration. Sensitivity analysis (on the
Problems were detected, proving the useful-
parameters used to calculate jbp) is the indicated
ness of the method, avoiding costly remedial op-
technique for this.
erations.
The field case studied here was concerned with
Field data analysis supports recording at
hook load measurement calibration - one of the
3 Hz, otherwise only a small part of the diagnosis
major sources of systematic errors. This was done
method could be applied.
by producing a pseudo friction factor log shown
in figure 22. NOMENCLATURE
The bottom curve was calculated using the
CTA values from the hook load analysis part. For J = binormal unit vector.
the other two curves, CTA was chmged by the CTA = center third average, Ibf.
amount shown on each curve. D = measured depth, ft.
As the& values show a trend line parallel to DWR = drag weight ratio, d-lessl.
e = error, d-less.
the depth coordinate and an average value of 0.37,
E = modulus of elasticity, psi.
there is no indication of any calibration problems.
For the purpose of describing the method, two 30 x 106 psi for steel.
f = mechan. friction factor, d-less.
other situations are explained (but are not taking
place here):
1) The & log for CTA shows slightly high
values at shallower depths decreasing as depth in- 1) d-less = dimensionless
170
WE Z(XWJ Jos& Veniincio Cardoso Jr.; Eric E, Maidl~ Luiz S. Idagawa 9
ACKNOWLEDGMENTS
The authors would like to thank Petrobr&
Unicamp and Fapesp. The authors also wish to
thank the Centre for Petroleum Engineering at
The University of New South Wales (in Australia)
for allowing the usage of all the communication
systems available to them. Also acknowledged
are the help of Mr. Yves de Malmazet, from
.
10 Problem Detection During ‘IMpping Operations In Horizontal and Dkectional Wells SPE 26330
K-MONEL I 8 3 29.7
STB 12110 e 3 5.1
r-
[
3 DC 101 Iblft
I
e 3/4
6
6
I
2 i3\16
3
4.7
I
I
92.3
1423
Figur4 2: Soratmlmclosurm.
-----
---------
---------
------
-----
,—.
I I
Figuro 6: Sddga
I
i
II
172
‘7 Figure
I f=rwrnal--
7 : Tripping
f=(ll
\
.
SPE 26330 JOS4 Ven&ncio Cardoso Jr.; Eric E. Maidla; Luiz S. Idagawa 11
E
I
I-
I
I
I
‘7
I
t
I
I
E
I
I
,
-41 r==-
Ml oooWav07 )iOOH Ml oooWavo7 WOH
. .. —- .
I I
t I
I I
I I
I I
I I
I 1
I t
t t
1 I
I I
I I
1 t
I I
I I
I I
t I
t !
1 I
I I
I
I
t
:.
I
I
I
I
I
1
%’
I
}
I
I
I
I
\ I
I
1 I
(Ft+Fn. &/.Ft
OWR =
/..; I F,
I
I
i
t
I
I
I
I
I
I
I
I
~
4
I 9 ‘w
I
F;.
I
1
1 PSEUDO FRICTION FACTOR CALCUfATION
1 .——.
1 ( 1101 valid fOf Wdh )
I
i 5i. Ft
lbp. fp .y
I
I n
I
I F. = lFt+F”. &+ eM,. IFt +FnibJ
I
I
I f@ - fbp + & + f~” O“L
1
I
I
I
I
I
Figure 13: Sketch for the example problem
I
I
I
WIl oooWavO~
X)OH
E sWOE
w
o <
~
a
PARTIAL
BOREHOLE
CLOSURE
Flgufo 14: Pewdo frkflcm fsclor Iypo cum. for parllal borahol@closuro
174
SPE 26330 Jos4 Ven&ncio Cardoso Jr.; Eric E. Maidlw Luiz S. Idagawa 13
[c OCEAN
~
>?
CATION
w II-l
n KEY SEAT r.
Q
w
z
5
Figure 16: Pseudo frlctlon factor type curve tor a key seat.
t
I
I
1
I
1
I
I
1
lJ*I -----------, I
I
,6 nm n. I
‘F’!’L
1
4 . . . . . . . . . .. . ..*m..
t
.’ *+#- I
~ “!.!
.’
~tw 4 . . . . .O”. ww I
I I
i
I
)
i u. .%
.’
“,,: *7 ,’
* m
hall 111,0
>! Iw!-.__—_ —U1
.. . .. .
JIM
170
~l,o
g
A
y 150
0
0
x
140 ~ = 0.3
..-&=o
13C
180
DWR = 5 %
170
:
~ 160
n“
< n
~ ------ ----- ----- ------ - ------------ J’t--
~ 150
0
g
. . ...4 =0.3
140
---&=o
130
, ,
</’
CALCUIAIED FOR CTA & 109
D 2530 2720