Professional Documents
Culture Documents
Complete J1or/e/iilg 01 Plunger lIlt CYcle A/lolYS Pror/uction OptimIzation 01 Intermittent Cas lIlt !fells
Although this combination of intermittent lift with plunger looks Rosina4 presented a hydrodynamic model of conventional plunger
potentially very attractive, it has received very little attention in the lift during the rising stage. In this work rigorous modeling of gas
published literature. Our literature survey showed, among other things: expansion from the annulus into the tubing, gas bubble rising at the
. No publications or case studies dealing with the effects of introducing same speed as the plunger, and plunger overruning the liquid slug
a plunger in an intermittent gas lift installation. according to experimental measurements, yield a set of ordinary
SPE Advanced Technology Series, Vol. 2, No. I 25
differential equations that can be solved numerically using standard Gas Blowdown (stage 3). After the liquid slug is produced and if
integration techniques. The results of this model compared well with operating conditions are such that the gas lift, valve remains open, a
experiments carried out in a 60 ft. high plexiglass experimental facility. third stage developes. This stage is characterized by gas flow from the
Liquid fallback behavior predicted by the model as a function of driving tubing-casing annulus into the production tubing. This is shown in
pressure ratio, was similar to that expected for an intermittent gas lift Figure 7. It has been reported by Brown 1U that under some operating
installation since bubble front and plunger are assumed to move at the conditions most of the production in intermittent gas lift wells is
same velocity. In other words, Rosina's plunger model will predict obtained from entrained liquid carried out by blowdown gas. In this
fallback curves similar to the one identified as "without plunger" in work this stage is included only to provide a rigurous transition
Figure 2. between stages 2 and 4 from the point of view of valve mechanics and
Mower and Lea. et. alb conducted a laboratory investigation on a gas consumption. As was the case during the previous two stages,
735 ft. deep instrumented well. This investigation provided pressure, reservoir inflow and gas injection from the surface are considered
velocity, gas and liquid volumetric information during the rise and fall during stage 3. If operating conditions are such that the gas lift valve
of 13 different commercial plungers. As a result, a substantial data closes concurrently with the plunger reaching the wellhead, stage 3
bank was completed with information that includes gas slippage, liquid does not occur and the model proceeds directly from stage 2 to stage 4.
fallback, and plunger rise and fall velocities. Figure 3 shows some of
the results obtained by Mower and Lea et al. For all plungers tested, Slug Regeneration/Gas pressure Buildup (Stage 4). As
the rate of liquid fallback was a linear a function of the average plunger illustrated in Figure 8, this stage starts when the gas lift valve closes.
rise velocity. This result will be used as an integral part of the model The plunger falls from the wellhead to the bumper spring at the bottom
to be developed in our investigation. of the well. Gas flows through the choke into the tubing-casing
Accordingly, the main objective of this paper is to develop a annulus, thus pressure bluids up. Fluids from the reservoir flow into
computer program to model wells produced using intermittent gas lift the wellbore, generating a new slug. Liquid left as fallback from the
with plunger. The model developed must model liquid fallback in such stages 1 and 2 also contribute to the volume of slug being generated.
a way as to reproduce observed experimental trends in the fallback After the plunger reaches the bottom and the tubing-cassing annulus
versus p,,/Pt ratio plot. A further objective is to stablish criteria for pressure reaches a certain value, the valve opens and the cycle is
production optimization of these wells based on sensitivity analysis complete. Thus stage 1 starts again.
performed with the program developed. The detailed development of the systems of non-linear coupled
ordinary differential equations for each stage, and the solution
MODEL DEVELOPMENT procedure, are presented in Reference 13.
The physical characteristics of an installation must be identified Althought it is agreed by most production engineers that the
before any attempt to model it can be undertaken. In Figure 4 we show efficiency of operation of any intermittent gas lift installation can be
an IGPL installation with all of the components being considered in this increased by adding a metallic plunger, no published work in this area
study. was found in the literature survey performed. In Figure 9 we show a
graph of predicted plunger velocity as a function of plunger position,
IGPL Cycle Stages for two liquid slugs of different mass. Note that plunger position is
measured from the bottom of the well.
In order to properly model the behavior of an IGPL installation, a The plungers experience very large initial accelerations, as soon as
complete operating cycle must be divided into phases or stages. Each the valve opens. Maximum velocities are reached 2 to 3 seconds after
stage is characterized by a particular combination of operating valve opening. These velocities decrease as the plungers move
conditions that distinguishes it from all other stages. upwards. This decrease can be explained, since as plungers move up,
Application of basic conservation of mass and linear momentwn the available pressure (at the bubble, just below the plungers)
laws to each stage yields a set of ordinary differential equations that decreases at a rate faster than the rate at which the slugs lose mass due
describe the dynamic behavior of all important variables during said- to fallback. Finally as the slugs reach the surface, the plunger
stage. velocities increase. This is also expected, since, due to a reduction of
In this work, we have divided an IGPL cycle into four different the weight of the slugs as part of the liquid enters the flowline, plunger
stages. velocities must increase rapidily.
As one would expect, the velocities experienced by the plunger
Plunger Rise (Stage 1). During this stage,. the gas lift valve is open pushing the heavier slug (846 ft long) are considerably smaller, than
and high pressure gas from the tubing-casing annulus pushes plunger the velocities of the same plunger pushing a lighter slug, other things
and slug in the upward direction. This is shown in Figure 5. being equal.
Fluids from the reservoir may flow into the wellbore accumulating Although plunger velocity is shown to vary, these variations are not
for the next cycle. Gas from the compression facilities flows into the large. Therefore the approach taken to account for fallback, as
tubing-casing annulus. Liquid fallback past the plunger is also described in the Reference 13, is justified.
considered, and is discussed in detail by Chacin et al l3 .
As a further point of interest, we show in Figure 10 the bubble front
Slug Production (stage 2). As soon as the slug reaches the wellhead, velocity versus position during the rise stage as obtained with the
liquid starts to flow into the flowline. This initiates stage 2, as shown in intermittent gas lift model of LiaoY This model, considers liquid
Figure 6. During this stage, slug and plunger velocities increase entrainment in the gas bubble and liquid film flow at the tubing wall.
rapidly. As was the case in stage 1, inflow of fluids from the reservoir, The bubble front velocity profile shows, as was observed with the
gas from the surface compression facilities, and liquid fallback occur IGPL model, large accelerations at the beginning and end ot the rise
during this phase and are considered in the model. stage.
26 SPE Advanced Technology Series, Vol. 2, No. I
For most of the rise stage, the slope of the velocity versus position There is another group of variables over which there is little choice
curve, however, is not negative, as in the IGPL case (Figure 9), but due to installation and cost constraints (P-rNT, P sep' dH' ~). Valve
shows an increase in bubble speed as the bubble moves upwards. type and plunger type are usually selected according to many criteria,
This behavior is expected, since in the case of intermittent lift without including availability, service, durability, etc. As a result in most
a plunger, fallback losses are larger. As a result, as the bubble front practical instances, the engineer will design the IGPL installation by
rises, the rate at which liquid mass is left behind is larger than the roecifying an apPfoviate combination of the following four parameters:
rate a which bubble front pressure decreases. The net result in the P d , R, 0 1, and 02' \
intermittent gas lift case is that the bubble front acceleration is The combined effect of varying these four variables can be
positive. presented in terms of total cycle time. By varying one or all of these
Clearly, then the qualitative behavior of our IGPL model is in parameters over practical ranges, the model developed can be used to
accordance with physical expectations. produce plots of total liquid production (BFPO) versus cycle time
In Figure 11, we present a plot of liquid fallback versus driving (min) and total gas injection required (MSCFO) versus time (min) as
pressure ratio P c!Pt for two different commercial plungers. The data shown in Figures 15 and 16, respectively.
for this plot was generated with the IGPL model developed. As shown in these two graphs, short cycles results in low
The shape of the fallback curves shows the right trend, with less production and high gas conswnption. This is the result of the
fallback observed in the low Pc!Pt ranges. As Pc!Pt increases, the reservoir only having enough time to build short slugs. Short slugs are
fallback losses approach a constant value. lifted at high velocities which in turn are responsible for larger
The two plungers compared in Figure 11 have similar fallback losses. Short cycle times are also responsible for a large
characteristics, except that plunger No. 21 has a hole thru its mid axis, nwnber of total cycles per day, which implies large gas conswnption.
and thus allows more gas to slip upwards, past the plunger, during the As cycle time is increased, the reservoir can produce longer slugs,
rise stages. It is observed in this figure that for a given Pc!Pt the which are lifted at slower velocities, with less associate percentage
liquid fallback losses obtained with plunger No. 21 are smaller than fallback. This translates into increased liquid production. Longer
those obtained with plunger No.3. cycle times also imply less nwnbers of cycles per day and, as a result,
As the reader may recall from the previous discussion on the work less gas conswnption per day.
of White and the results presented in Figure 2, the results obtained As shown in Figure 15, there is a cycle time at which maximwn
with the IGLP model developed and presented in Figure 11 are in good production time is obtained. A further increase in cycle time results in
qualitative agreement with experimental evidence. To the best of the a decrease in total production. This is because, although longer times
author's knowledge, this is the first time that a fallback versus Pc!Pt imply longer slugs, slower rise velocities and less fallback, they also
curve obtained with any plunger model, reproduces the trends imply less cycles per day which must ultimately reduce total liquid
observed and reported by White. This fact further suppors the production on a per day basis.
implementation of fallback based on average plunger rise speed, as As shown in Figure 16, gas injection requirement always decreases
presented in this work. as cycle time is increased. This has an important implication: unless
Although it is agreed by most production engineers that the the price of gas and compression costs are negligible, the optimwn
efficiency of operation of intermittent gas lift installations can be cycle is not necessarily that at which maximwn liquid production is
increased by adding a metallic plunger, a survey of 15 oil production obtained.
companies worldwide showed that no data for IGPL installations is The IGPL model developed can be used effectively to generate
available. IGPL installations have been very seldom used in Venezuela curves similar to those shown in Figure 15 and 16, which are great
also. It was only possible to fmd complete and reliable data for one tools to select optimwn cycle times, based on maximum liquid
well (Lagoven's PB-616). Figures 12 and 13 show model simulation production or maximum profit.
results for this well plotted along with the measured field data (oil
production, gas conswnption and total cycle time). The agreement is
quite good with differences among the measured values and the Optimization of Existing Installations
simulation results being less than 5%. One of the important parameters that control cycle time in IGPL
installations is surface choke diameter setting, O? From the point of
MODEL IMPLEMENTATION view of practical applications, choke diameter setting is even more
important, since it can be easily adjusted by the operator at the
In this section we will discuss how the model developed can be welhead, without any need to perform workover operations or stop
used effectively by production engineers under three different production.
schemes: design, optimization and artificial lift method selection. Ideally it is desirable to determine the effect of choke setting on
liquid production and gas injection requirement. This can be easily
Design ofInstallations. done with the help of the model developed. Figures 17 and 18 show
In Figure 14 we show schematically an IGPL installation. The plots of liquid production (BFPO) and gas conswnption (MSCFD)
parameters identified in this figure are needed to fully characterize the versus surface choke sitting for 3 different dome pressures, as
behavior of any field installation. The model developed includes all obtained with the IGPL model. As expected, these plots show that
of these parameters and thus allows one to study the effect of varying if the choke diameter used is too large, gas injection flow rate is
these parameters on well performance. unnecessarily high, resulting in short cycles, short slugs, and large
From the point of view of an engineer trying to design an IGPL fallback. The opposite occurs if the choke diameters selected is too
installation for a given existing well, the following well variables are samall: low injection flow rates, long cycles, long slugs and not
fixed: PRES' PI, GOR, gAPI, ZRES' dc, TWH' TRES' LH' dH' enough cycles on a per day basis to maintain an optimwn production.
The term fixed is used to imply that these are given well variables, An interesting point to reflect upon by observing Figures 17 and 18
characteristics of the well under study. These variables however can is the effect that dome pressure has as a parameter. In field
be changed in the model in order to simulate any well under operations,
consideration.
Symbol Description
CONCLUSIONS
WH Wellhead
H Flowline
The results obtained with the model developed in the present work C Casing
fully support the following conclusions: t Tubing
- A new model has been developed for the design of plunger lift TC Tubing-casing annulus
assisted intermittent gas lift installations. TCI TC, upstream of valve port
- The model incorporates all important production system components. TC2 TC, downstream of surface choke
and includes reservoir inflow performance characteristics and gas lift 1 at valve
behavior. 2 at surface choke
- The model treats fallback estimation in a novel way, through the use BSL Bottom of slug
of laboratory derived liquid slippage cha,racteristic of commercial TSL Top of slug
plungers. B Bubble
- Model predictions agree qualitatively with observed behavior in PL Plunger
SL Slug
plunger lift installations. The model compared within 5% with data
BRO Tubing, at valve level
obtained in one well from Western Venezuela.
RES Reservoir
28 SPE Advanced Technology Series, Vol. 2, No.1
REFERENCES
10. Brown, K.: Gas Lift Theory and Practice. Prentice Hall, Inc.
(1967).
13. Chacfn, J., Schmidt, Z., and Doty, D.: " Supplement to SPE 23683,
Modeling and Optimization of Plunger Lift Assisted Intermittent
Gas Lift Installations," Paper SPE 28162 available from SPE,
Richardson, Tx.
Autnors
Jesus Chacln is Production Methods Section Head at INTEVEP, SA,
where he has worked in artificial lilt since 1987. Dr. Chacln holds B.S. and
M.S degrees in Mechanical Engineering from Rensselaer Polytechnic
Institute and a PhD degree in MechanICal Engineering from from the
University of Oklahoma. Zellmlr Schmidt is Professor of Petroleum
Engineering at the University of Tulsa and Director of Tulsa University
Artificial UltProjects. He spent ten years as production engineer wi1h
INA-Naltalin IOYugoslavia, and has served as a consultant to various
companies. Dr Schmidt holds an Er19ineering degree from the University of
Zagreb, M.S. and PhD d~rees 10 Petroleum Engineering from the
University of Tulsa, and is a Registered Engineer in Oklahoma In 1987-88,
he was SPE's distinguished lecturer. Dale R. Doty is Associate Professor
of Mathematical Sciences at the University of Tulsa and Associal8 diredDr
of Tulsa University Artificial Ult Projects. He joined the faaJlty in 1975, and
has been involved in both research and consulting in the area of Petroleum
Production. He holds as, MS, and PhD degrees in Mathematics from
Michigan State University. (SPE 2M])
SPE Advanced Technology Series, Vol. 2, No.1 29
ADJUSTABLE
CHOKE
I 50·~~,------------------r=~~
"-
........
WITHOUT PLUNGER
CATCHER 30
I----ASSEMBLY BRUSH CAPILLARY
~:::SO==L~ID~)~~=====l!!!,..:ot
-..-
........
PRODUCTION
.....
20
10
\
- ~ .... ~ CAPILLARY (HOLE)
- CONTROLLER
o '· . . . . . GROOVED
BALL
VALVE MOTOR VALVE
1.2 1.3 1. 1.5 1.6 1.7 1.8 1.9 2
t-;=~--- GAS Pc I Pt
b SOURCE
REDUCED FALLBACK
ZONE FOR PLUNGER Pc : CASING PRESSURE (psI)
LIFT Pt : TUBING PRESSURE (psi)
S: INITIAL SLUG LENGTH (It)
D: DEPTH OF PLUNGER STOP (It)
RESERVOIR
WELUIEAD
1 ---n
7------I--J
SEPARATOR
0.08
0.07
CHOKE
0.06 TUBING
0.02
No.1 GAS LIFT VALVE
0.Q1
PACKER
0
0 10 20 30 40
D2 PINJ
ZBsL
ZREs Zp ZRES Zp
1
ZL ZL
[01. R)
PSEP
PSFJ>
I PWH
02 PINJ
:J
ZPL
ZRES Zp
ZREs zp
1
TICI
PTeI
1
a I [01. R)
~
AI: 846 ft .Iug
J' ...l
~
20
~~
10
~ 10
0
0
!!:
~ 0
0 2000 4000 6000 8000
~
0 2000 4000 6000 8000
PLUNGER POSITION (I'l) == BUBBLE FRONT POSITION POSITION (ft)
FlGURE 9 PLUNGER RISE VELOCITY PROFlLE ACCORDING TO MODEL FlGURE 10 BUBBLE FRONT RISE VELOCITY PROFILE ACCORDING TO
DEVELOPED INTERMITTENT GAS LIFT MODEL OF LlAO
1400
1300
20 WELLPB .616
1200
~
:
0 1100 • = SIMULATION
/ ®=
#3 (VALVE ROD) 1000
e FIELD DATA
:Ill ~ 900
U
« 10
~
I
# 21 CAPILLARY (HOLE)
800
~ 700
6 600
~ 500
~ C
400
~
::s 0
300
200
2 3 4 100
Pc I Pt 0
0 2345678
FlGURE 11 FALLBACK VS DRIVING PRESSURE RATIO CURVES FOR
TWO DIFFERENT PLUNGERS ACCORDING TO MODEL DEVELOPED CYCLE TIME (MIN)
FlGURE 12 GAS REQUIREMENT VS CYCLE TIME
150
140 WELLPB ·616
130
120 • = SIMULATION
~ 110 ®= FIELD DATA
c 100
Z 90
g 80
U 70
;J 60
~ ~g
30
20
10
O+-~--~~--~-r--~-r--~-r~
o 20 40 60 80 100
CYCLE TIME (MIN)
FIGURE 13 PRODUCTION VS CYCLE TIME
~
§ 70
f
r--+--+--- cl 60
0 10 20 30 40
r-t--- cl;
TOTAL CYCLE TIME (mm.)
7JUlS ZP
FIGURE 15 LIOUID PRODUCTION VS TOTAL CYCLE TIME
ACCORDING TO MODEL DEVELOPED
200~---------------~
eo 60
p
§
f
50
40
0 0.1
Pd =1300 psi
0.2 0.3
c;-
.e-
e
80
60
IGPL =
PR 100 PSI :l
all ~JGPI.
SURFACE CHOKE SIZE (In.) Z PR=400 PSI
0 40
FIGURE 17 LISUID PRODUCTION VS SURFACE CHOKE !=
SIZESEW~orn~J¥~~:l~~~~gsuRES, g
Q 20
f 0
1000 0 100 200 300
~ 400
~
fIl
200
c-<
0
0 0.1 0.2 0.3
SURFACE CHOKE SIZE (In.)
(SPE 23683)
FIGURE 18 GAS INJECTION VS SURFACE CHOKE SIZE SETTING
FOR 3 DIFFERENT DOME PRESSURES, ACCORDING TO MODEL
DEVELOPED