Sunteți pe pagina 1din 12

pcbtyc+mrdsm-sl

SPE 27975
Practical Well Test Interpretation
Louis Mattar, Fekete Assocs. Inc.
SPE Member

-ht

Errginaere,Inc.
1~, Wc@tyofPetroleum

Thk paper wee preparadfor ~ntetion

at the Universityof Tuks CentennialPetroleumEngineeringSymposiumheld in Tulsa, OK, U.S.A., 2S-31 Auguef1S94.

This paper wee eefectedfor p&smation by an SPE ProgramCamStea folkwing review of informatknoontalnadin an ebetrsofeubmlffedby the author(e).Contenteof the paper,
se W-M,
hsve not been reviewedby lhe SoolefyOf Pefrokum Enginsateend are subjectto comotion by the author(a).The material,se preesntad,does not neweserlly refkot
any positionofthe SoMatyof PetrofeumEngineere,ireoffkera, or membem.Peperepreatmfedet SPE meetingsare wbjecf to pubfksfionreviewby EdlforlalOommineesof the society
Of~fdellm En@naara.P~
tOc@y je ~
to en abatrectof notmorethenWOwords.Illuetretione
maynotbe copied.The sbatrecfaftoufdwnfein conapbwouseoknowfadgment
of where and by whomthe papa lapresented.Write Ltbrarisn,SPE, P.O. Sox S3SSSS,Richerdeon,TX 750SS.3SSS,U.S.A. Telex, 1S3245SPEUT.

Abstract
Wa!! Test !nterpreta?!on !nvdves much more than
Pressure Transient Analysis. It includes scrutinking
the field notes, the surface operations, the equipment
and the wellbore configuration. It means incorporating
geological information and production information;
adjusting the measured data to reservoir conditions
and accounting for multiphase effects both in the
wellbore and in the reservoic recognizingthat wellbore
effects during a buildup are differentfrom those during
a drawdown; in short, integrating the practical issues
with the theoretical analysis.
FieJd examples will be presented to illustrate these
. . A .- **Q
effects. :gncirinM.-e-h ~n !ead tO the wrong
pressuretransient analysis.

Introduction
Reservoirengineering integrates many looksinto the
resewoir; for example core analysis, log analysis,
pressure analysis and production analysis. Each one
of these gives its own view into the reswvoir. Some
only investigate a few inches (core, logs) while others
reflect a much larger (100s of feet) radius of
investigation (pressure, production). The lookscan
be supportive of each other and fit in with other looks
into the reservoir, such as geology or seismic
interpretations. However, quite often, these looks
can be contradictory of each other, in which case
175

some of the ev~ence must be rejected, ignored or


explained away. It is the responsibilityof the amlyst to
integrateall these perspectivesof the resetvoir in order
to arrive at the best poesibie itit&pie*&tk3rk
The same analogy applies within each discipline. For
example in the field of pressure analysis, Pressure
Transient Analysis (P.T.A.) must be viewed as being
only part of the picture, not the whole picture. Well
Test Interpretation (VV.T.I.)must encompass the field
notes describing how the test was conducted,
previous tests, initiil and final static gradient, the
wellbore configuration, multi-phase aspects in the
wellbore, wellbore dynamics (references 1, 2 and 3),
material balance and production information, in
addition to P.T.A.
P.T.A. has become a very sophisticated tool, and
because of@ advancednature, it tends to be placed
on a pedestal, to the neglect of other more useful yet
simpter Information. For example, a P.T.A. on a flow
and builduptest may indicate a dual porosityreservoir.
While this may be the case (and there are many
wellbore
dynamics that may contra-indicate that), we
may have overbooked the fact that static gradients
before and after the test indicate depletion (a nonm AII
k$~ue
u~, I nnrQ~&
~-.
economic reservoir - Ihance Lh~
becomes a red herring).
This article is not intended to belittle the value of
P.T.A., rather it attempts to emphasize the point that
this powerful tool must not be used in isolation, and
that itsvalue is greatly enhanced when it is used within

PRACTICAL WELL TEST INTERPRETATION

the ~r~der

persp@ive of We!! Test Interpretation

(W.T.I.). In the rest of the paper, I will discuss several


cases (unrelated to each other) each illustratingsome
practicalaspect of wefttesting.

1. Measure Initial Pressure


Figure 1 represents synthetic data of a buildup test
(150 hr. flow 31 hr. shut-in) as published in the late Dr.
The dtiiculty, as
Rame~s paper (reference 4).
expressed in that publication, was that there was no
correct Homer straight line, and at best the
permeability that could be derived from this Horner
plot was 37.6 mD as compared to the true permeability
of 48 mD.
In this particular example, the initial pressure was
known but was not displayed on the Horner plot. If we
place the initial pressure on the x-axis = 1 (infhite
shut-intime) as in Figure 2, it becomes evidentthat the
pressure is returning to the initial pressure. If the
semifogstraightline [s anchored on the initialpressure,
then the value of permeability calculated is 44.8 mD.
This is significantlybetter than the previous calculation
and is acceptably close to the true answer. [If you
know where you have come from, it is easier to know
where you are going to.]

SPE 27975

gauge or a strain gauge. Mattar et al (reference 1) give


an example ~- strain gtwge and a quart! gauge from
different manufacturers which track each other
flawlessly. Figure 5 shows data from a strain gauge
and a (supposedly much more accurate) quartz
gauge; the one is building up while the other is
dedining; the difference between them is some 200
kPa, and independent measurements show the strain
gauge to be the correct one. ~he quality of service is
more importantthan the type of the instrument.]

4. Is TemDemture lmDortant or Not?


Electronic pressure gauges are known to be
temperature sensitiie, quartz much more than strain
gauges. Thus it is importantto have a good handle on
the temperature when measuring the pressure,
otherwisesignificantinterpretationerrors can occur.
Figures 6 and 7 show two different tests, each using
two strdn gauges by the same manufacturers. Both
tests show a temperature anomaly. In Figure 6 the
pressures from both recorders track excellently in
spite of the temperature discrepancy. In Figure 7 there
is a temperature anomaly of the same magnitude, yet
the pressuresdo not track, diverging at early time and
converging at late time. [Sometimes it matters and
sometimesnot.]

2. Desian Your Test


5. Freauencv of Samoling
In this case history,there were 6 oil wells in a pod, one
in the centre and five in a circle around it. In order to
establish
htervvell
properties
and
confirm
communication, a puise test was corxkmtsd with the
central well as the active well. Out of the fwe
observationwells, two (2) responded as expected from
the design (Figure 3), two (2) showed a very confused
response (Figure 4), and one (1) showed no response
at all to the active well, but seemed to respond to a
well producing from another pod!
~hings dont
abvaysgo the way you plan them.]

Unlike analog gauges, digital electronic gauges do


not provide a continuous pressure trace, but they give
individualdata points at each sample time. In order to
optimize the number of data points, a sophisticated
sampling program can be designed, whereby a large
number of pressures are measured when there are
rate changes, from one rate to another, or from flow to
shut-in (or vice-versa), and fewer data points are taken
elsewhere. It has often been observed that a change
in the frequency of sampling can result in a change in
tk recorded pressuretrend as illustrated in Figure 8.
This is obviously not a reservoireffect but the artifact
of the electrorks of the pressure recorder. Aiso the
more complex the sampling program is, the more ft is
prone to operator error. Moreover, having a complex
Pm-programmed sampling scheme removes a lot of
the flexibilityneeded during some tests. For example,
if the test had been designed to flow for 12 hours and
then be shut-in, but haffway through the flow period,
the well started loading up or hydrating and it became
necessary to modifythe test, the complex pre-set

3. Select Your Pressure


Gauae Semite Commmy
A lot of emphasis is placed on selecting the pressure
gauge for a particular test. Often it is more important
to select the right service company than to select the
right gauge. Even the best of gauges need to be
properly calibrated and maintained. The question
often arises as to whether one should use a quartz
176

SPE 27975

L. MATTAR

sampling frequency program may be totally


inappropriate for the modified test, and much needed
valuable data may be missed as a result. [Keep it
simple.]

for example, what is the true reservoir pressure? did


the injection create a hydraulic fmcture which heals
during the falloff? nhe P.T.A. models may be too
ideaii%d to representa real-lifesituation.]

6. Wellbore Tubular Confiauretion

9. Mvetarious Date

Figure 9 shows data from two downhoie recorders. it


is obvious that the more you flow, the higher the
pressure gets! This stmnge observation is readily
that
the
recorder
is
explained if it is recognized
situated
some90 m below the sliding sfeeve, through
which gas is flowing from the annuius to the Wb!ng.
The space between the sflding sleeve and
the
recorder run depth is slowly filling up with water
throughout the test. If the pressure at the end of the
test is corrected for this hydrostatic column of water,
then the final reservoir pressure is some 300 I@a less
than the initial reservoir pressure, indicating severe
depletion during the test (in spite of the apparent
increasing pressure). [Conduct a static gradient
before and after a test, and account for tubular
configurations.]

!n S*Q 9f ail effofts, we were unable to find an


explanationfor the periodicityobserved on the biiiidup
test of a dry gas weii show-n in Figwe %3. Both
electronic pressure recorders showed the same
oscillations. (We may never know all the answers.]

7. Rermat&ble Anomalies
Figure 10 is a plot showing two buildup tests
conducted on the same OIL well 6 months apart. They
both show a significant non-resewoir anomaly (60
kPa) some 7 hours into the test. At the end of the test
a static gmdient run in the well showed GAS and not
OIL to be present in the wellbore. The cause of the
anomaly is obviously a wellbore dynamic associated
with phase segregation and liquid efftux. [Wellbore
dynamics can be very obvious, biif they maybe subtie
and inconspicuous.]

8. Drawdown (Injetilon) versus


Buildug fFalioW]
P,T.A. is based on the dmwdown (Injection) equation
and Buildup (Falloff) analysis utilizes this solution
along with the principJeof superposition. In theory, the
analysis of drawdown (injection) data should yield the
same answers as buildup (fafloff)data. In pmctice, this
happens only mrely. Very often, the analysisfrom the
flow period gives different resuftsfrom the analysis of
the shut-in period. Figures 11 and 12 are examples of
significantly different analyses of the injection and
falloff data on the same test. Such inconsistencies
must be resolved by information external to the test;

10. Non-Uniaue Solutions:


Many well tests can be interpreted using several
different reservoir models. Selecting the appropriate
model often cannot be done without recourse to
external information eg. geological deposition.@
environment. Figures 14 and 15 show 2 ditYerent
modefs simulating the same test data with equally
acceptable results.
In THEORY, every reservoiris unique, and its behavior
However, in
is different from other reservoir.
PRACTICE, the differences between many reservoir
models are so small that they are marked by the
scatter of practicaldata measurements. In the event of
the multiplicity of models, the late H.J. Rame~s
philosophy must be heeded, namely [start with a
simple model and proceed to a progressively more
complex model, only as you need to.]

Conclusions
(1)

If you torture the data long enough, it will


confessto anything.

(2)

Pressure Tmnsient Analysis (P.T.A.) should be


utifized witinin the broader %arnework of Well
Test Interpretation(W.T.L).

(3)

W.T.!. >> P.T.A.

PRACTICAL WELL TEST INTERPRETATION

FM~iWiCW3

1. MATTAR, L

CriticalEvaluation and Processing


of Data Prior to Pressure Transient Anaiysis,
Presented at the 67th Annual Techrkal
Conference and Exhibition of the Society of
Petroteum Engineers, Washington, B.C., October
4-7, 1992.

2. MAITAR, L and SANTO, M.S,: How Wellbore


Dynamics Affect Pressure Transient Analysis,
The Journal of Canadian Petroleum Technology,
Vol 31, No. 2, February, 1992.
K.: ~b
pd~v
3. MAITAR, L and ZAO~
Pressure Derivative (PPD) - A New Diagnostic
Tool in Well Test Interpretation,The Journal of
Canadian Petroleum Technology, VOI 31, No. 4,
April, 1992.
4. RAMEY, H.J.: Advances in Practical Well Test
Analysis,J.P.T., June 1992.

178

SPE 27975

PRESSURE

HORNER
BUILDUP

PLOT

3200

2800-

2400

.IJl
? 2000

A
A

G1

2
1600

A
A
A
A

1200

A
I,i!,

1,,!!
102
(t+At)/At

Iii

800
,04

103

PRESSURE

HORNER
BUILDUP

,Ql

1.0

Figure 1

PLOT

3200

Anchor
onPi
Pi

A
A
A
A
A
A

1600

A
A
A

1200I

I
800F
,~4

I,,l.
103

1,,,!
102
(t+At)/At

179

Ii)!!
,Ql

1.0

Figure 2

Observation

Weii Response (weiis 1 and 2)

pmour. at

*
,
,

0
~

~
a

10100 -

,
,

10050 =

10000

,,,

(2.0 120
Figure 3

Observation

Weii Response (weiis 3 and 4)

.. . . . . ..- -------------

10200

~
10150
.
0
10100

,
10050

~;

29

::,,

Time

.(hrj

Figure 4
180

Comparison of Quartz and Strain Gauge

14800

k
Quartz

______

,.

..-

------

-------

------

-----

-----

--

-1

-k
I

i
L

i
14000

Figure S

inconsistent Temperature

but Consistent Pressure

18000~771
. ....

*
&.--=?..
i

. --------

-------

#..
No,,.
*=tir-e
pr**..--

/
/

-- --=-+.-in

12000

,
/
/~

~
/~.,J

70

-----

*c~..

15000

--

----

_
Gr-e*o
P

-rg

~69

-/,/

..;
68

9000

.-

#(---

, $ 0

Ld-w

,/0

/9
~/ #
(4

1
6000

\
67

/h4
**S
,~
,
18
/
I
,
w,,,,,,,,,,
l=..0

~(-)~o ...

~~

t
t

100

105

~ 110

115
..- -----

#z

/.-.

I ftm,~r,r

125

130

. .......... .1
.
,4066
135

Figure 6

181

Inconsistent Temperature

- Inconsistent pressure

4501

6690

165

r
6660

6570 ;

-------

------------------

Pressure
Difference
> =-----

141

~
654051

1
/
/
51.5

52.5

52

53

53.5

Time,

54

----------------------54.5

55.5

55

5.5

o 1]

60

(hr)

FigWO 7

Effect

of Sampling Frequency on Pressure

--J

14450

14448

14446

.. ..

,..

... ....

SiiilingFrM mcy,
<=
=:3 minute

/
\

-:.
.... .....

~--14442

\:

\:

L
.2

.4

my......

=~ minutes
i: -

..... .!
... .Y.........
..................1... ,,,...%&,
:,/
$, ;
- /<
Y
... ~{
.......... .....
....................... ...........?+

*,:\
*:-\\

/
I

.. .....

......................... TN... .. +a@iu9.R$Q


/
ft

14444

...
_
.
1
.1.

...........+#f
.
ix
/

191.2
Time, (hr)

191.4

- .19

&%

.8

192

Figure 8

162

Apparent

Increasing Reservoir

Pressures Attributed

to

Tubular Configuration

-------n

3600

:
5.
a-1

o
IJ
:

2400

20000.0

520

40

Time

360

400

.(hr)

Figure 9

Two BuiIdup Tests, 6 months apart

Tim

(ha-

163

>

Figure 10

Inconsistent Fall-off

Model

2700
92 4.0
0.34
-0.361

DI k=
s=

2400

21OC

.m
a 180(
a

150(

1201

A,.l.-,..l,,,,
',,,,`''`''`'``'''''
'''``'''''15'```'`''~'
o'`''''`''''`'L
21
18
12
9
6
3
o

90

24

Figure 11

t,hr

Ininetinn
,. .,-----

Ir
pl

2400

Sloo
C,. o

k=
s=

4.

0.022
-4.790

18
15

l,!! ....ll . .. . . ...l .. .... l


5

MQdQl

12
t,hr

184

21

1!,...
24

Recordei
Pws-%fo,kpa

#2
#1

,Recorder

two
tom

Sz

f.lw

6$
00

00

00
00

00

Pressure,
(kPa)-------Pressure,(kPa)
(mm
mm
C)o
C)o

4
(
(
(

I
.
c
c

-+

0
....................................
.:

0 :

....................................
0 :

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

G
o

0 :

...
...=:
...;......}.
2

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

N :
-1
-.

~i

o
cd
N

u
~ L...

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

:):

o:

,:

o
0

. .

1 ,

Wn

I,,t,tlt.ttlttttltl,

for..)
Glu

;.-----------(3
(~

6aP)tiL
&Jsl
6ep).

........1.:
........

Lu.l
I/lu

(AA
. .

Inw

al

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

1#
z#

**
44

JaPJO~e>l
JapJ03a&j-

U!(JI
Uxn

COMPOS

,04

MODEL

/~

1(33

paaters
kl -

V
c:

u-

0.

100.00
5.000
100.00

104.49
47.18

I 1-E

rl=

&

k2

I
al

10:

,01
,0-2

:J

~~d~--
,()-1

,01

1.0
At

102

, L

1(

Figure

15

S-ar putea să vă placă și