Conditions
In Sucker Rod Pumping Wells
S. G. GIBBS
SHELL DEVELOPMENT CO,
MEM&9f AIMI?
A. B. NEELY SHEIL 011 CO.
ASSOCIATE MEMBER AIME HOUSTON, TEK.
.
‘-- {
.
1
in which the & are separation constants. Anticipating
Fourier series expansions in t, assume periodic solutions u(d) = -!$ + 5 v. cosn@l-&sin tit . . (6)
~=1
$ILE$l ~ as T(t)= e’””’ from which the ~. can be evalu-
- -a. + i/3., n = 1,2,3, ..., wherein: In. order that the polished rod displacement condition be
satisfied, the choice of constants stemming from L@Jt) =
Re [z(OJ) ] is made. This results in:
T(r):~ T
X(x):qx, g cs.
=+-J U(IAt)sin n~tidt , n = 1,2, . . . .
0
where & q and ~ are real constants.
To satisfy the load and displacement boundary condi- As before, the above integrals are evaluated numerically
tions, the product solutions chosen arw by harmonic analysis methods,
Whh the preliminary work of evaluating the constants
of integration @mpleted, the equations giving subsurface
2(x,?) = gqx -t & + 5 (% sin Lx -1- displacement and load may be developed. The formula
*=1
@. cos &x)e’”Wt ,, . , , . . . . . (4) for displacement at arbitrary depth and time U(XJ) is
,
bbtained by evaluating the real part of Eq, 4. This results
where iw.
% = –x. – ifin ,
‘“ + : o.(x) Cosmot +,
u(x,t) = — .,2> ,
,2;Ax+~ ..l
@.=v. —i8n .
P.(x)sinnd, . . . . . . . . (7)
The load signals which are received at the surface can be
represented by truncated Fourier series of the type where O,(x) = (K. cosh /3.x + S. sinh /l.x) sin LY.X
+ (p. sinh /3.x + v. cosh /3tix)cos a.x,
D(4 = ~ + ~~lu. cos nmf+r. sin twt . . , (5)
and P.(x) = (K. sinh @ -1-8,, cosh /3.x) cos a.x
Referring to Eq. 4, the load condition can be satisffed by
taking . - @. cosh ~.x -t v. sinh /3.x) sin ax.
F(xJ), the dynamic load at arbitrary depth and time,
““”=‘e[EA-l > n may be obtained from Eq. 7 and I-Iooke’s law: F(x,t) =
which leads to a choice of constants given by: EA au(x’r) These considerations yield:
T
u. = .?M(K&Y.
+ pnp.) , n =.1,2, “““,
To= Ez4(K.& — PKI.] , n =’1,2; . - i,
~(,x,t) = EA & -1- 5 on’(x) Cos no)t +
[ fi=l
2EA&7 ,
1
u. =
Pn’(x)sinnd , , , . . . . . . (8)
where the u and T are Fourier cocficients associated with
the load curve measured at the polished rod, These co- where
efficients are evaluated using harmonic analysis, i.e., by
numerical evaluation of the integrals
0.’(x] = “&,sinh /3.x + @J3. - v.a.) cosh &x “sin a.-x
T [
rrn=~ D(@ COS mid , tt = 0,1,2, -,-,
T J
o
-T
+ [ &
—cosh ~nX i- (I@ + &Y,) SiIIh @nX COS %x,
1 a.
and
— —________ &L=-+__._!!?@ @?n_@@_ _L&~!2?Z.~.”L .._
–- --&,(;i g- --’r,-. . . . .
J
o
~co%’ pnx + (K/X = Gi)-siiih-pk “-i35iGZz’”‘ ‘——
[ 1
‘l%e displacement of the pdiskd rod vs time is also a
boundary condition which must be satisfied. This condi- — -!ksinh pnx + (w& i- &a.) cosh ~~x sin U.X.
tion is representable by the truncated Fourier series: [ EA 1
9s JOURNAL OF PETROLEU?II TECENOLOGX
.. . . . . .. . . .. _-. ,. . . . . . .. . ... - ..”. “e. -. ... . . . ..- . . ..— . .._. . . . . . .. .
,-.
. ..—.
E@. “7 and 8 are the basic relations from which subsur- 5 and 6). Ordinates from these curves for one complete
face displacements and loads are deduced from data mea- cycle are r~d into the computer along with ,the pertinent
sured at the surface, IsI a sense, these are the equations rod design daa pumping speed, damping factor, etc., to
which permit the decoding of the subsw!face data which yield the set of dynagrapli cards shown in Fig. 2. Consid-
has been transmitted through the sucker rod. T’hese equa- erable diagnostic information can be gained from analysis
tions can be easily generalized to represent the displace- of these cards. Shown are the calculated dynagraph cards
ments and loads in wells which have any number of at the surface, at junction points in the combination rod
changes in rod cross-sectional area. string and at the pump. Peak stresses of 28,S00 psi in
the l-in. rods, 29,600 psi in the ?6-in. rods and 30,300
A COMPLETE EXAMPLE psi in ~the %-in, rods are indicated. This reveals a good
rod desi~ in which peak stresses are reasonably balanced
The primary aim of a dynamometer analysis is to mea- in each rod interval. The pump dynagraph card indicates
sure loads and displacements at the polished rod and, from a gross pump stroke of 7.1 ft, a net liquid stroke of 4,6
these measurements, deduce down-hale conditions. The ft and a fluid load of 3,200 lb. On the basis of the
distinctive feature of the analytical technique described gross and net strokes, a liquid volumetric etilciency of
here is the manner in which surface data are interpreted. 65 per cent is calculated. From the fluid load and pump
Rather than relying solely on visual interpretatio~ which size, an intake pressure of 690 psi is estimated. The
is dependent on personal judgment and experience, the shape of the pump card indicates some down-hole gas
technique uses a digital computer to calculate subsurface compression. The shape also indicates that the tubhg an-
cards, thus furnishing quantitative results from which the chor is holding properly. A liquid displacement rate of
personal element has been largely removed. 200 B/D is calculated which, when compared with mea-
To illustrate the technique, the polished rod data of sured production of 184 B/D indicates no serious tuling
Pig, 1 is analyzed. These data are shown in the time or flowline leak.
history form as measured by a special dynamometer and This illustration indicates that a great deal of quanti-
pertain to an t3,525-ft well having a three-taper rod string tative down-hole information can be gained from a single
operated with a 74-in. conventional unit at 15.4 strokd computer analysis. The whole’ operating status of a pump-
min. Measurement of polished rod data in time history ing well can be diagnosed and summarized in a set of cal-
form is most convenient in view of the need for express- culated dynagraph cards (Fig. 2). The example is particu-
ing these data in Fourier seriez as functions of time (Eqs. larly appropriate because of the odd shape of the polished
rod card. This is a graphic case in which visual inter-
26.00. ~
pretation of the surface data would be difficult, because
the pump dynagraph card in no way resembles the sur-
face card.
r
4000
. . ...= . .. ___ -.. -.— —--- —.. —.. —-— .—— —.-+ -.. . .—
–:7000
4 3 2 -1 -3 -4
PUhiP LilS:LACEMENT, -f:
r
8000
F
ed the annulus to the flowline was defective, and part of
the production was being recycled down the casing. Re-
pair of this valve increased the production rate to 140
B/D. This type of malfunction is fairly common and has
been encountered on severaI occasions in applications of
the technique.
-2000
I I I I 1 1 I J
DETECTION OF UNANCHORED TUBING 3 z ,
PUMP 01S?LAGE61 CM-;, f f
-z -3 -4
Defective tubing anchors sire frequently not discovered m PARbi-Ltt06W4 w-mm wdP Mm m wm row UfJbfKKIRED ruwm
ing affects the manner in which the fluid load is trans- I% 5—EFFECTOF Tumixc MOVEMENT
ONPunrr CASDSXAFE
ferred to and from the rod string. This is manifested in
the polished rod data, where it can be detected with the release are not always as straight as those in Fig. 4, Also,
analytical technique. In the mathematical development the by the stretching and contracting of the unanchorcd tub-
displacements of the pump are measured with respect to ing during the period when the fluid load is bein’g trans-
a coordinate system fixed in the ground. If the tubing is ferred to and from the tubing. The sequence of events
unanchored and moves relative to the ground, then the shown in Fig. 4 depicts the movement of the pump and
pump dynagraph card referred to a tubing coordinate the shape of the pump card relative to both coordinate
system will ha~ a dKferent shape from the card referred systems for a high-efficiency well which has negligible free
to the ground coordinate system, For a pump dynagraph gas passing through the pump.
card which is rectangular with respect to the tubing (saY, The sequence in Fig. 4 has been idealized for clarity.
a card measured with the Gilbert dynagraplf which rides It should be noted that the tubing is an elastic body
along with the tubing) the technique will indicate a paral- which, when unanchored, vibrates during intermittent ap-
lelogram-shaped card. The parallelogram shape is mused plication of the fluid load .Thus, in cards calculated with
preferences tiven at end of paper. respect to ground, the segments during load pickup and
( /“”8’”
@ @ @
I
PIMP HAS !?EACHEO LOAD TRANSFERAL EN7cZL FL”’U LOAD
“lGMES? PC$N7 fiND rROLt mm 10 S.,P.WED B, T“BAG
$ 0CSCENC8VG. FLUID T“B’\o &LW05T 7“9 h$ MM s? RETCWD
Lou Is mt!o mANs- co!fPLcrE Tumo w tin., cwmsr
FE.WtO FROX RODS CON18N”ES 70 WW3T” P“UP ,s DE-
10 1uB6NG. TuWK IS STRETCH. Scmom, 6 m WI Ms.fa
Slnclc”tldt. Pull? !5 10 601. GRVWD ANO
DESCCf10’N& m, mm 7“wW COOiW,liATC
TO G80W0 W tS SYSTEMS.
51&Tt0NBRV Wil “
RE6PCCT ‘IO T*
Tlmho.
10&ma
. i- u.. e F
.—-. —..— . . .._ ._ ._.
--i-4-li===7--”---
the fluid load is quickly applied; therefore, the tubing pumping well is its intake pressure. There are several
stretch indicated by the calculated card will usually ex- methods for determining pump intake pressqre, both direct
ceed the stretch calculated on the basis of a statically or and indirect, Direct methods include use of a pressure
SIOWIy app]jed fluid load. An cxample is shown in Fig. 5a bomb which. is lowered down the annulus, and various
for a well with 4,oOO ft of unanchored 2Y2 -in. tubing. electrical transducers which transmit pressure data up from
The parallelogram shape is evident. the pump via electrical cables strapped to the tubing.
In wells having effective anchors, the tubing is re- When the direct methods are not stymied by inadequate
strained from moving. As a result, pump cards are the annular clearances or cabling troubles, they are the best
same with respect to both coordinate systems. Fig. 5b and most precise methods to use. The most common in-
shows a calculated pump card in “a deep well in which direct method involves measurement of annular fluid lev-
“a tubing anchor is installed. The rectangular shape of the els by acoustical means. Once the fluid level is known,
card indicates that the anchor is operating properly, the intake pressure is calculated by assuming an effective
It can be seen that the method of detecting tubing move- gradient for the fluid in the annulus. However, if the an-
ment just described becomes less conclusive in wells which nular clearance is inadequate, acoustical measurement of
have considerable free gas passing through the pump. In the fluid level may be impossible, and intake pressure esti-
these wells the expansiori-compression effects of the free mates in this manner may not be obtainable,
gas can be confused with tubing movement. Fortunately, Another indirect method for estimating intake pressure
it is frequently possible to distinguish between tubklg is provided by the analytical technique described herein,
movement and gas expansion-compression etiects by the This method is not hampered by space limitations or elec-
shape of the pump card. Good pump spacing minimizes trical cabling troubles, because all down-hole data are
gas expansion effects during rod load pickup; hence, ex- transmitted through the sucker rods. Though not capable
cessive movement of the pump during this period can imp- of great precision in determination of pressures, the meth-
ly tubing movement. od can be used to obtain engineering estimates of intake
pressure which oftentimes are adequate.
It is interesting to note that use of the analytical meth-
od for detection of tubing movement was not anticipated Determination of pump intake pressure by the ana-
but grew out of attempts to understand why some com- lytical technique is based on the equation:
puted pump cards had a parallelogram shape rather than
the expected rectangular shape. P,= Pr, –~. . . ..L. . . (9)
v
CHARACTERISTIC DYNAGRAPH CARDS The meaning of this equation is illustrated in Fig. 7. The
FOR LEAKY PUMPS term P, denotes pressure above the pump caused by the
In a number of instances, badly leaking pumps have head of fluid in the tubing. The terms W, and A,, denote,
been indicated by calculated subsurface cards. When a respectively, fluid load and pump area. Their quotient.
pump is leaking due to a bad traveling valve or plunger, W,/AP denotes the, pressure ditlerence across the pump.
it is slow to pick up the fluid load at the beginning of the Thus, the intake pressure P, (the pressure beneath the
upstroke and tends to release the load prematurely before pump) is found by subtracting the pressure difference
the downstroke begins. The load pickup is slow because
plunger velocity is low, and the leak rate is almost as
large as the plunger displacement rate, Near the end of
the upstroke the plunger velocity is again low, and the
plunger displacement rate becomes smaller than the leak
rate. Thus, the load transferal begins while the pump is
still ascending, These effects give the pump card a-n arched
top. A typical calculated pump card which indicates a 1 I I I
leaky pump is shown in Fig, 6a. In this card the slow load .1 0 -1
Fufd. DrS.PLACE!/E N,, F,
pickup and premature release are evident, A1l pumps
[al QYNAGRAPH ctm FOR LEAKING PLW.GER CR TRAVELING VALVE.
leak to a certain extent, but when the leak becomes large
enough to give an arched top to the pump card, remedial
action is usually warranted.
Fig. 6b shows a calculated pump card for both a leaky
pump and an unanchored tubing string. On load pickup
the effects of tubing-shortening and the leaky pump are
additive, Near the top of the stroke the effects of tubing
stretch and the leaking pump cancel. Thus, the parallelo-
gram shape of the unanchored-tubing card combines with -looo~,
the arched-top shape of the leaky-pump card to yield a P~14P DIs PLfi CEMENT, f!
-~’b ---
pump card an arched bottom.
The diagnosis of leaking pumps in field tests has been
-–—-contlrmed-by- mechanical--inspection- after-fhepurnp-was––
-2 -1 0 -3 -4
ESTIMATION OF ~PUMP INTAKE PRESSURE PUMP D15PLACEUENT, f!
One of the most valuable diagnostic ciues about a FIG.6-TYPICAL DYNAGRAPIiCARDSFoa LBAKINGPUMPS.
JANUARY, 1966 9s
. .....-_ ______ -. .._ -. ...,. ._ ----- ,, ----------- . . . .._ .- .._. . . . . . .. . .. .... .... . . . . . . ... ... . ------ . _-_,----- ___
... . .... .. . . . . .. . .. .. . . . . . ..
. .
—. -.. .
across the pump W1/AP from the pressure above it PD.The pumped under packers.’ In such we[ls, fluid levels cannot
‘ fluid load WI is read directiy from we calculated pump be obtained because of the packer in the casing-tubing
dynagraph card. The pressure above the pump may be annulus,
inferred from gradient considerations or from a two-
phase-flow pressure traverse. Determination of pressure DETECTION OF GAS SEPARATION PROBLEMS
over the pump is the primary factor which limits the ac-
curacy of this indwect method, A sucker rod pump is a relatively poor gas compres-
To illustrate the use of Eq, 9 in estimating intake pres- sor. Consequently, it is desirabie to minimize the free gas
sure, the data of Figs. 1 and 2 are anaiyzed, From the which the pump must handie because this gas decreases
calculated pump card, a fluid load WI of 3,200 lb is de- the volume in the pump available for liquid, causing a
termined. For the 1.5-in. pump in die well (AP = 1.67 diminution in the pump volumetric efficiency. To provide
sq in.), the pressure d~erence across the pump is W~/AP adequate down-hole separation so that free gas, passing
= 1,810 psi. The pressure above the pump P. is deter- through the pump will be minimized, it is desirable to
mined from a pressure traverse as being approximately know the severity of any existing down-hole gas separa-
2,500 psi. FinaIiy, according to Eq. 9, the pump intake tion problems. The shape of the pump dynagraph card is
pressure is estimated to be P, = 2,500- 1,S10 ==690 psi. one of the best intilcators of such problems,
Fortunately, it is not necessary in many cases to know As the analytical technique provides the shape of the
the intake pressure with great accuracy, Much about the pump card, the severity of down-hole gas separation prob-
potential proctti tivity of the well can be inferred from lems is readily apparent. A pump card which is generaliy
the shape of the calculated pump card. A well having a rectangular (Fig. 8a) indicates a negligible amount of free
rectangular-shaped pump card is not pumped off-regard- gas passing through the pump. The rectangular shape
less of its intake pressure. The PVT properties of the pro- “ arises because the fluid in the pump chamber is virtually
duced fluid can also be of use. For example, a pump incompressible;, hence, the load transferal at the top and
card which indicates gas compression signifies that the in- bottom of the stroke occurs with negligible movement of
take pressure is below the bubble point pressure, which the pump. If there is free gas in the purrsp, on the other
is known from the PVT properties of the crude. hand, the fluid in the chamber is not incompressible and
Another useful procedure is to calci.rlate an upper limit the pump does not remain stationary during ioad trans-
@ the intakb pressure. This is done with Eq. 9, in which feral. Pig. 8b ,shows a typical pump card which indicates
the pressure over the pump is computed on the assump- compression and expansion segments for the entrained gas.
tion that all of the fluid in the tubing is ~il and water in The distahce that the pump travels during the expansion
the proportions indicated by a weli test. It is then known and compression phases is lost stroke insofar as liquid
that the actual intake pressure is. less than this upper iim- production is concerned. The expansion curve at the be-
it, because any gas in the tubing acts to decrease the ginning of the upstroke is strongly affected by pump spac-
prewure over the pump. It is tiseful to note that the ana- ing. To minimize stroke loss due to gas expansion, the
lytical method provides the oniy known means for de- pump should be spaced with the minimum” practical clkar-
termining intake pressures in tubingkss completion wells, ance at the bottom of the stroke.
In these wells there is no annulus for determination of a
fluid level, nor is there any tubing on which to attach EFFECT OF EXCESSIVE ROP FRICTION
electrical cabl+ for pressure transducers. The technique
can also be used to estimate pressurea in wells being As previously mentioned, a convenient way of regard-
ing the sucker rod analytical technique is by analogy with
a communication line. Data indicative of down-hole op-
A. erating condhions are received at the polished rod after
&l&z traveling up from the pump along the sucker rod. Ordi-
I narily, the steel sucker rod transmits these data with great
TUBING HEAD PRESSURE
fidelity. However, if the data are somehow garbled en
route, they may “not be suitable for computer anaiysis. In-
deed, the oniy conceivable way for the technique to fail is
\ for data to be transformed and distorted while traveling
through the sucker rods. A factor which can create this
\
“m
— t
g FL._iIdu 54
[0) Nf.OLIBIBLE
3210
.PvUP
FREC
DISPLACEMENT,
GAS IN PUMP
-1-2-3
ft
-4-5
~o
54 3 -2 -s -4 -5
P&!P D: SPLAC:MBN;,’ ft
PRESSU8E
(b) M00CR6TE FREE !lAS IN PUMP
FIG.7%SCIIEMATICOF DEPTSi.PRIISSURE
RELATIONSHIP
IN
PUMPINCW%LL, FIG. 8--EFFECT OF FSLSE
GAS ONPuiwP CARD SHAPE.
garbkg effect is excessive friction on the rods caused by the well load depends on the torque factor which in tum
crooked holes or corkscrewed tubing. depends on the geometry of the pumping unit. Normally,
Striking proof of the adverse effect of excessive rod the torque factor is determined by a solution which in-
friction is shown in Fig. 9a. These cards were calculated volves all of the linkage dimensions of the unit. However,
in a well in which a packer was used for a tnbhg anchor. if the assumption of constant crank angular velocity is
The calculated pump card is unrealistic and uninterpre&- made, the simple relationship between poiished rod veloc-
able. When thh card was calculated, it was surmised that ity and torque factor can be derived:’
rod friction garbling was occurring. Later, when the well
was pulled, severe rod coupling wear was noted. This in- TF= ~ . . . . . . . . . . .. (10)
dicated that the tubing was corkscrewed by its own weight U
as it rested on the packer. When the well was placed back
on production, it was equipped with a tension-type tubing This relationship is particularly well suited to the method
anchor. Fig. 9b shows the polished rod card and pump described here because the polished rod velocity is easily
card after the tension anchor was installed. As indicated, derived from the displacement function (Eq, 6) already
surface loads are less severe and the calculated pump card required for computation of subsurface conditions. Thus,
is reasonable. the torque factor may be determined from Eqs. 6 and 10
When the analytical technique is correctly applied with as
reliabIe data, the only way that it can produce an incor-
rect pump card is for excessive rod tilction to be present. TF = j –tzv. sin mot -f- rd. cos not . . (11)
Therefore, an uninterpretable and unrealistic pump card ~= ~
frequently implies adverse rod friction which should be
corrected if possible, The complete formula for net torque can now be written
W
DETERMINATION OF TORSIONAL LOADING Net Torque = TF[D(.]t) + W,] – C sin (d + ~), (12)
The applications discussed so far primarily concern the from which a torsional analysis can be made. A typical
determination of subsurface operating conditions. Aiso of torque analysis is presented in Fig. 10. This pertains to a
great importance is loading on the surface equipment, par- 4,724-ft well being produced with a 168-in. unit operating
ticularly torsional Ioading on the pumping unit gearbox. at 9.5, strokes/rein. Fig. 10a shows computed dynagraph
It is therefore desirable to be able to determine both sub- cards and Fig, 10b shows corresponding net torque and
surface and surf ace operating conditions” in a single com- counterbalance torque curves developed from Eq, 12, As
puter analysis. A brief description is given which illustrates indicated, the unit is reasonably well counterbalanced with
how polished rod data required for computation of sub- the peak torsional loading being about 520,000 in.-lb.
surface Conditions c-m also b: used to deter@ne gearbox
torques at the surface.
The net torque on the gearbox is composed of the
torque caused by the well load, and the opposing torque
created by the counterweights. ,The torque arising from
25,000
.20,000
n
~- I 5,000
u
0
~ I 0,000
a lb
s
~
-1 6000
I \ PUMP CARO \
o.
1 I I I I I
4 3 2 -3 -4
Put!P OIS:LACE;~NT , %
[0) WELL EOUIPPEO WITH SET - 00WN PACKER
. . .,-,.
25,000
r
.$ x+ /-
n
z
~
5000
PUMP CARD
2
o Lf --
-. ---