Professional Documents
Culture Documents
Jafar Ov 2016
Jafar Ov 2016
This paper was prepared for presentation at the Offshore Technology Conference Asia held in Kuala Lumpur, Malaysia, 22–25 March 2016.
This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Offshore Technology Conference, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the
written consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words;
illustrations may not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.
Abstract
Producing gas wells faces with the challenge of liquid loading mostly in later stages of production. Due
to formation of liquid coloumn in the wellbore, the rate decreases below the critical rate. Hence, water
reduces well performance. Depending on the wells, liquid coloumn can form because of either entrance
of formation water to the wellbore in dry gas wells or wet gas liquidification in wet gas wells or
condensation in gas-condensate wells. In all the cases, as a result, bottomhole pressure increases and
reservoir pressure does not have enough pressure to force the liquid to the surface. This leads to
considerable reduction of well production and that may lead into seizure flow in more severe cases. In last
decades, different techniques and correlations have been offered by many authors for determination of
critical liquid loading rate. This paper reviews the existing several widely used liquid loading models and
highlights their advantages and drawbacks. Additionally, at the end of the paper, a case study has been
conducted for these models to identify their applicability conditions and limitations regarding the selection
of best matched based on well data. According to the case study’s results, Guo’s model predicts critical
flow rate more accurately than all other models. There’s no any applicability ranges, which it can be
applied to all vertical gas wells with different wellhead pressures. When it comes to the other models, their
applicability ranges should be taken into account carefully for not to ending with incorrect estimations.
Introduction
Initially, shallow gas wells have low flowing bottomhole pressure. When liquid introduces itself in the
wellbore, bottomhole pressure starts to increase. Sequentially, gas velocity decreases and reaches its
critical rate, which below this rate liquid loading process expands in the wellbore and the flow converts
from unloaded to loaded regime. Fig. 1 schematically illustrates the liquid loading process. As seen from
the figure, liquid rate in gas flow increases from mist to bubble by decreasing gas velocity. The gas well
can enter to one or all of these regimes of its life. Consequently, tubing filled with the liquid increases
bottomhole pressure and reduces production and ultimate recovery.
Observation of onset of liquid loading in most cases is very difficult, because loaded well may continue
to producing for long period without any symptoms. It can be determined through several ways in later
stages of production life of the well. Liquid loading can be easily observed through Nodal analysis by
plotting IPR & OPR curves. In Fig. 2, the red point shows that the well is already loaded and prevention
2 OTC-26526-MS
Figure 1—Liquid Loading Steps in the Well (Lea and Nickens 2004)
methods must be considered to remove liquid from the wellbore and make the well healthy (green point).
Nodal analysis clearly shows that liquid loading problem is very important for gas wells and immediate
actions should be taken as early as possible. Fig. 3 is an example for gas production rate fluctuations. It
is observed from the figure that after a stable production, the rate begins to fluctuate. This is the first
symptom of liquid loading.
Figure 3—Detected liquid loading from production data (Sutton et al. 2003)
Many different models have been developed to deal with this problem and forecast future performance
of gas wells. All these models are either based on theory or empirically based on experiments. Some of
the developed models might not be implemented in the industry because of their limitations. The
following part of the paper reviews these models from past to current.
Literature Review
where kv is 1.3.
The velocity that can suspend the largest liquid droplet at the wellhead is called ⬙Turner critical loading
velocity⬙. Turner et al. reported that out of 106 wells, 16 are questionable. 66 wells out of 90 best fitted
with their model. Then they adjusted the equation 20% upward and got 77 wells best fitted. Note that
Turner et al. used 30 as a Weber number.
(3)
(4)
However, since Turner’s data sets were mostly consisting of high wellhead pressures (Pwh⬎1000 psia),
they could not determine that 20% adjustment gives incorrect results for low wellhead pressure wells.
Another drawback of Turner’s model was an assumption that the shape of droplet is spherical and does
not change while flowing in the well.
Coleman et al. (1991) firstly reported that Turner’s 20% adjustment does not work with low-rate and
low wellhead pressure wells and should be applied without adjustment. Coleman’s non-adjusted droplet
model is widely used for wells with wellhead pressures less than 500 psia.
Nosseir et al. (2000) by working on Turner’s data set, announced that Turner’s equation is not based
on the flow regimes and this leads to some errors in the calculations. They calculated all Reynolds
numbers for each rate and concluded that Turner’s assumption for Reynolds numbers between
104⬍Re⬍2⫻105 is not correct. Most of the data exceeds this range and shows itself in highly turbulent
regime. As a result, they proposed two new equations regarding the critical velocity: one for transition and
one for highly turbulent flow regime.
For transition regime
(5)
(6)
Li et al. (2002) unlike Turner’s spherical-droplet model, introduced new flat-shaped droplet model.
They explained that in a high velocity gas flow, the fore and apt portions of droplet have a pressure
difference. This pressure difference changes droplet’s shape from spherical to like a convex bean form,
which has unequal sides. Fig. 5 shows the droplet’s shape changes from spherical to flat in a high velocity.
Compared with spherical droplets, flat ones need low gas velocity and flow rate due to having more
efficient area. For the Reynolds number range 104⬍Re⬍2⫻105, drag coefficient (CD) for Turner’s model
is 0.44, but for flat shaped one is 1.0, which means smaller critical velocity than spherical droplet.
(7)
(8)
Figure 5—Entrained Droplet’s Shape in a High Velocity Gas Stream (Li et al. 2002)
Although, Li et al. introduced the new shaped droplet model and showed Turner’s spherical droplet
model is not completely accurate, they made the correlation based on the assumption that the shape of
droplet is constant and does not convert. They did not consider if droplets coalesce, how it can change
their shape. In reality, experiments show that droplets coalesce and separate many times in the wellbore.
Consequently, they will need more gas velocity to move upward.
Guo et al. (2006) found that Turner’s method is not accurate, because they considered only top of the
wellbore instead of bottomhole conditions in their model. However, since the flow in the wellbore is
multiphase, bottomhole condition should dominate rather than top hole and because of it their assumption
results with the incorrect minimum flow rate prediction. Based on this observation, Guo et al. introduced
new phenomena - minimum kinetic energy criterion for calculation of critical rate for multiphase flow in
gas wells. They showed that gas should exceed the minimum kinetic energy value in order to transport
the liquid droplet to the wellhead. On the other hand, multiphase flow model predicts accurate bottomhole
pressure and fluid density that are used in kinetic energy calculation. They proposed a minimum required
gas production rate calculation based on this new model:
(9)
(10)
6 OTC-26526-MS
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
The calculations are followed through eqns. (10) – (22) and got final ␥ value. This value is compared
with the value of eqn. (9) after substituting the values of eqns. (10) - (22) in (9). The calculations are
repeated until right hand side of eqn. (9) is close to ␥ value of eqn. (14). Eventually, calculated Qgm
becomes minimum flow rate that gas transports the droplet to the surface.
Wang and Liu (2007) offered that for Reynold number ranges between 104 and 106 and Morton
number ranges between 10-10 and 10-12 in gas wells. They determined that most of the droplets are
disk-shaped in gas flow and may be carried to the surface because of having more efficient area similar
like flat-shaped one. They also calculated that drag coefficient (CD) for disk-shaped droplet is 1.17 and
proposed new equations for critical velocity and flow rate:
(23)
(24)
OTC-26526-MS 7
Same as Li et al. model, they also did not consider in their model in case of droplets coalesce and
separate and how it can change the minimum flow rate.
Zhou and Yuan (2009) showed that for gas wells gas velocity usually is very high and the flow is
turbulent. In turbulent flow regime, droplets do not flow just through upward, they flow through all the
directions. Droplets may coalesce and make bigger ones, then may separate to small ones and then again
they may coalesce. This process is repeated continuously. The process is illustrated in Fig. 6. From this
observation, Zhou and Yuan introduced a new definition - liquid holdup to represent liquid droplet
concentration in gas wells:
(25)
where Hl - liquid holdup and sl and sg are superficial liquid and gas velocities, respectively.
They explained that conventional Turner’s critical velocity equation is fitted, if liquid holdup is equal
or less than threshold value of liquid-droplet consentration () Hlⱕ.
(26)
However, if Hl⬎
(27)
where vcrit.–N - critical velocity for new model, ␣ is a fitting constant. From Turner et al (1969) data,
␣ ⫽ 0.6,  ⫽ 0.01 were estimated. The maximum Hl ⫽ 0.24. According to Barnea et al. (1987), when
the liquid holdup becomes higher than 0.24, the multiphase flow changes to slug flow pattern.
Luan and He (2012) by analyzing both Turner et al. (1969) and Li et al (2002) results, determined that
Turner’s model overestimates, whilst Li’s model underestimates the critical velocity of gas wells.
Consequently, they introduced a new dimensionless parameter loss factor - S for their new model to
include the gas energy loss caused by the move of flat-shaped droplets of Li’s model. Thus, the empirical
equation of model
(28)
where vcrit.–T and vcrit.–L are the calculated velocities from Turner’s and Li’s models, respectively.
8 OTC-26526-MS
Luan and He analyzed production data of 300 gas wells in China and observed that S factor ranges
between 0.75 and 0.83. For simplification as an upper limit 0.83 is used.
(29)
(30)
Authors, recommended to use their model with low-pressure gas wells, especially wellhead pressures
less than 500 psia as Coleman’s model.
Case Study
Case Description
A case study has been performed to identify the applicability conditions and limitations of the models
based on Turner’s well data set and provide a deep insight. The critical flow rates have been calculated
based on the application of different models for 68 out of 106 wells reported by Turner et al. The well
parameters have been given in Table 1. The reason not to do the calculations for all the wells is the lack
of some well data in Turner’s original set and being unwilling to make assumptions for them, as it could
result in errors. In addition, the reported 16 questionable wells have not been considered as well, because
of their incorrect well parameters, which even though the calculations can be performed through them. On
the other hand, average z factor has been used for simplifying the calculations. Using an average value
doesn’t have any noticeable impact, because they are closed to each other.
Results
Table 2 summarizes the critical gas flow rate calculations for all the models. Fig. 7 through 13 show the
results for the case study in calculating the critical gas flow rates for the wells. All figures have been
divided by blue line to two regions. The right side identifies loaded, while left side identifies unloaded
regions, whilst blue line itself is near loaded-up region. Fig. 7 shows Turner’s original liquid loading
prediction after 20% adjustment. Turner’s correlation shows that it is good for unloaded wells with
wellhead pressures higher than 500 psi. As Coleman reported, 20% adjustment gives incorrect predicted
rates. Since it overestimates, this correlation cannot be used for already loaded wells. Analysis of
10 OTC-26526-MS
Nosseir’s model in Fig. 8 shows that this model works best for the near loaded-up wells with wellhead
pressures less than 1000 psi. The model gave incorrect minimum flow rates for 10 wells out of total 34, which
is a quite high error as a percentage. Investigation of Li’s Model in Fig. 9 displays that unlike Turner’s model
it underestimates the rates for mostly loaded wells. The correlation gave overall 12 incorrect predicted rate
values, which it is due to consideration that droplets are only flat-shaped as explained above. Li’s rates for
unloaded wells are close to Turner’s ones. Review of Guo’s Model in Fig. 10 shows that their correlation
predicts the critical rates correctly almost for all the wells. Although, in the figure 6 incorrect rates have been
shown, 5 of them are close to near-loaded up regime. Guo’s model predicted only 1 well’s critical flow rate
completely incorrectly, which is result acceptable. The results prove that not just wellhead, also bottomhole
conditions in the same time should be considered in estimating the minimum critical flow rate of wells to get
correct values. Similar to Li’s model Wang’s model also underestimates the critical flow rates as shown in Fig.
11. The number of incorrect predicted rates even increases to 17 by the application of Wang’s model. This is
because of an assumption that droplets are only disk-shaped as explained above. Analysis of Zhou’s model in
Fig. 12 shows that introduction of the liquid holdup phenomena increases the accuracy of critical rate
estimation. Furthermore, it is understood that Zhou’s model predicts loaded wells better than unloaded ones.
However, the main poor side of this correlation is its overestimation character. Because of this reason, 11 wells
have been predicted incorrectly. Introduction of the loss factor in Luan’s model, gives more reasonable critical
rate values than Turner, Li and Wang’s Models as displayed in Fig. 13. Compared with those three models (Fig.
7, Fig. 9 and Fig. 11), Luan’s model did only 7 wrong predictions. Additionally, plotting the critical velocity
values against the wellhead pressures graphically proves that Turner’s model overestimates. It also shows that
Li and Wang underestimate.
Table 2—Calculated Critical Flow Rates by Different Models Based on Turner’s Data Set
Turner’s Turner’s
Observed Droplet Adj. Nosseir’s Li’s Guo’s Wang’s Zhou’s Luan’s
Test Rate, Model, Model, Model, Model, Model, Model, Model, Model,
Pwh, psi Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day
725 775 779 935 813 542 880 381 956 841
400 417 583 700 503 270 – 190 579 630
108 568 306 367 264 147 243 104 316 330
540 712 661 793 690 313 500 220 807 714
450 442 419 503 361 286 470 201 421 453
3607 1525 1156 1387 1207 778 1349 547 3936 1248
3434 2926 1150 1380 1671 761 1333 535 3683 1242
3660 3726 1142 1370 – 783 1396 550 3978 1233
3340 2611 2412 2894 2518 1690 3033 1188 17555 2605
3540 1814 1635 1962 1707 1155 2069 811 11627 1766
3525 1792 1108 1330 1157 770 1401 541 7404 1197
3472 2572 1085 1302 1675 765 1411 537 7229 1172
3338 2261 1623 1948 1695 1125 2039 790 10855 1753
3092 3351 1574 1889 – 1086 2001 763 9626 1700
3455 2769 1082 1298 6459 763 1411 536 7054 1169
3665 2542 1660 1992 1733 1173 2052 824 8714 1793
3615 3890 1648 1978 6184 1166 2057 819 8534 1780
3212 2547 1604 1925 1675 1105 1919 777 6251 1732
3025 3517 1569 1883 – 1075 1877 756 5746 1695
8215 3472 1956 2347 2042 1629 2951 1144 8860 2112
7405 6946 1930 2316 – 1570 2810 1103 7738 2084
2226 1959 910 1092 – 623 1036 438 1696 983
1590 3009 3281 3937 3426 2089 3424 1468 3661 3543
1520 4150 3195 3834 3772 2044 3351 1437 3584 3451
OTC-26526-MS 11
Table 2 (Continued)—Calculated Critical Flow Rates by Different Models Based on Turner’s Data Set
Turner’s Turner’s
Observed Droplet Adj. Nosseir’s Li’s Guo’s Wang’s Zhou’s Luan’s
Test Rate, Model, Model, Model, Model, Model, Model, Model, Model,
Pwh, psi Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day Mscf/day
Figure 7—Calculated Critical Flow Rate Based on Turner’s Adjusted Model vs. Test Flow Rate
Figure 8 —Calculated Critical Flow Rate Based on Nosseir’s Model vs. Test Flow Rate
OTC-26526-MS 13
Figure 9 —Calculated Critical Flow Rate Based on Li’s Model vs. Test Flow Rate
Figure 10 —Calculated Critical Flow Rate Based on Guo’s Model vs. Test Flow Rate
14 OTC-26526-MS
Figure 11—Calculated Critical Flow Rate Based on Wang’s Model vs. Test Flow Rate
Figure 12—Calculated Critical Flow Rate Based on Zhou’s Model vs. Test Flow Rate
OTC-26526-MS 15
Figure 13—Calculated Critical Flow Rate Based on Luan’s Model vs. Test Flow Rate
Figure 14 —Critical Velocity Comparisons among Turner, Luan, Wang and Li’s models
Table 3 summarizes the results of the case study. Doing comparisons through all models concludes that
Guo’s model predicts critical gas rates more accurately and gives overall reasonable results. These results
prove that bottomhole conditions have more effects than wellhead conditions in critical rate calculations.
Conclusion
The following conclusions are made:
● Liquid loading is one the most important problems for gas wells, which immediate curavative or
preventive actions should be taken.
● Prediction models for critical flow rate of vertical gas wells have been studied through extensive
literature survey and explained.
● Case study has been performed for the critical flow rate prediction through the different models
and described their advantages and limitations to get more deep understanding.
● It is recommended to use several different models in the same time, while determining critical gas
velocity and rate. Making comparisons between the models, increases the accuracy of calculations.
● This study can also be further broadened for deviated and horizontal wells individually or together.
Nomenclature
vcrit. ⫽critical velocity, ft/sec
⫽surface tension of liquid to gas, dynes/cm
l ⫽liquid density, lbm/ft3
g ⫽gas density, lbm/ft3
CD ⫽drag coefficient (recommended value 0.44)
Qcrit. ⫽critical flow rate, Mscf/day
A ⫽cross-sectional area of flow, ft2
p ⫽pressure, psia
z ⫽gas compressibility factor
T ⫽temperature, °F
p ⫽liquid drop density, lbm/ft3
⫽gas density, lbm/ft3
Ai ⫽cross-sectional area of conduit, in.2
Dh ⫽hydraulic diameter, in.
Ekm ⫽minimum kinetic energy required to transport liquid drops, lbf-ft/ft3
f ⫽Moody friction factor
L ⫽conduit length, ft
phf ⫽wellhead flowing pressure, psia
Qgm ⫽minimum gas flow rate required to transport liquid drops, Mscf/day
QG ⫽gas production rate, scf/day
Qo ⫽oil production rate, bbl/day
Qs ⫽solid production rate, ft3/day
Qw ⫽water production rate, bbl/day
Sg ⫽specific gravity of gas, air ⫽ 1
So ⫽specific gravity of produced oil, fresh water ⫽ 1
Ss ⫽specific gravity of produced sloid, fresh water ⫽ 1
Sw ⫽specific gravity of produced water, fresh water ⫽ 1
Tav ⫽average temperature, °F
Tbh ⫽bottomhole temperature, °F
’ ⫽pipe-wall roughness, in.
⫽inclination angle, degrees
Hl ⫽liquid holdup at wellhead
Vsl ⫽superficial liquid velocity, ft/sec
OTC-26526-MS 17
References
Turner, R.G., Hubbard, M.G., Dukler, A.E. 1969. Analysis and Prediction of Minimum Flow Rate for the Continuous
Removal of Liquids from Gas Wells. J. Pet. Tech. 21 (11): 1475–1482
Coleman, S.B., Clay, H.B., McCurdy, D.G., Norris, L.H.III 1991. Applying Gas-Well Load-Up Technology. J. Pet. Tech.
43 (3): 344 –349
Libson, T.N., Henry, J.R. 1980. Case Histories: Identification of and Remedial Action for Liquid Loading in Gas Wells
– Intermediate Shelf Gas Play. J. Pet. Tech. 32 (4): 685–693
Nosseir, M.A., Darwich, T.A., Sayyouh, M.H., El Sallaly, M. 2000. A New Approach for Accurate Prediction of Loading
in Gas Wells under Different Flowing Conditions. SPE Prod. & Fac. 15 (4): 241–246.
Li, M., Li, S.L., Sun, L.T. 2002. New View on Continuous-Removal Liquids from Gas Wells. SPE Prod. & Fac. 17
(1):42–46.
Lea, J.F., Nickens, H.V. 2004. Solving Gas-Well Liquid-Loading Problems. J. Pet. Tech. 56 (4): 30 –36
Lea, J., Nickens, H.V., Wells, M. Gas Well Deliquification, 2nd Edition. Elseiver, Gulf Drilling Guides 2008.
Guo, B., Ghalambor, A., Xu, C. 2006. A Systematic Approach to Predicting Liquid Loading in Gas Wells. SPE Prod. &
Oper. 21 (1): 81–88.
Guo, B., Ghalambor, A. Natural Gas Engineering Handbook, 2nd Edition. Gulf Publishing Company, Houston Texas
2005.
Solomon, F.A., Falcone, G., Teodoriu, C. 2008. Critical Review of Existing Solutions to Predict and Model Liquid
Loading in Gas Wells. SPE Paper 115933 presented at the SPE Annual Technical Conference and Exhibition, 21-24
September, Denver, Colorado, USA.
Zhou, D., Yuan, H. 2009. A New Model for Predicting Gas-Well Liquid Loading. SPE Prod. & Oper. 25 (2): 172–181.
Luan, G., He, S. 2012. A New Model for the Accurate Prediction of Liquid Loading in Low-Pressure Gas Wells. J. Can.
Pet. Tech. 51 (6): 493–498
Riza, M.F., Hasan, A.R., Kabir, C.S. 2014. A Pragmatic Approach to Understanding Liquid Loading in Gas Wells. SPE
Paper 170583 presented at the SPE Annual Technical Conference and Exhibition, 27-29 October, Amsterdam,
Netherland