You are on page 1of 14

Prediction of water coning and water cresting: analytical or

numerical models?

F. Verga, D. Viberti, P. Ferraro, Politecnico di Torino.

This paper was presented at the Offshore Mediterranean Conference and Exhibition in Ravenna, Italy, March 28-30, 2007. It
was selected for presentation by the OMC 2007 Programme Committee following review of information contained in the abstract
submitted by the authors. The Paper as presented at OMC 2007 has not been reviewed by the Programme Committee.

ABSTRACT

In the presence of an active aquifer, water coning (vertical wells) and water cresting
(horizontal wells) phenomena may cause a significant reduction of the well productivity.
Water coning phenomena affect the efficiency of the oil drainage from the reservoir, thus
reducing the ultimate oil recovery, increasing the cost of production operations, and arising
environmental problems related to water disposal. After water breakthrough at a well, the
water phase may become the dominant one, and the well must be shut in because oil
production becomes uneconomical.
Historically several efforts were concentrated on controlling water coning. A wide number of
analytical and semi-analytical approaches were developed to analyse coning phenomena
and to define a valid methodology to evaluate the critical rate. The critical rate is the
maximum oil rate that can be produced before water breakthrough occurs. However, due to
the complexity of the problem, a rigorous approach of general validity could not be identified.
The present study was aimed at providing an exhaustive comparison among the most used
analytical methods and the results obtained by numerical simulation for water coning and
water cresting prediction for a number of possible scenarios. In particular, the results
provided by the analytical and numerical models were validated by comparison against
production data of some real cases.
In general, in the case of vertical wells the comparison showed that the critical rate obtained
from the analytical methods underestimates the critical rate obtained from numerical models,
especially for high anisotropy ratios and low oil viscosities. On the opposite, in the case of
horizontal wells the critical rate obtained from analytical methods strongly overestimates the
critical rate obtained from numerical models. The errors in predicting the critical rate for
vertical wells with analytical models with respect to numerical simulation can be acceptable
in most cases, whereas a reliable evaluation of the critical rate for horizontal wells should
only rely on numerical modeling.

INTRODUCTION

Water coning is a serious problem in many oil fields because it causes a significant reduction
of well productivity. According to Hamed (2000) the phenomenon of water coning is the result
of movement of reservoir fluids in the direction of least resistance balanced by the tendency
of the fluids to maintain gravity equilibrium.
One of the main reasons for coning is the pressure drawdown induced by production. While
a vertical well exhibits a large pressure drawdown near the well bore, horizontal wells show
minimum pressure drawdown so that coning tendencies can be minimized and higher oil
production rates can be achieved. Production from the well creates pressure gradients that
tend to raise the oil-water contact in the immediate vicinity of the well. Three different forces
may affect fluid flow distribution around the well bores. Firstly, capillary forces have an

1
influence but their effect is negligible. Secondly, gravity forces have a strong influence in the
vertical direction and arise from fluid density differences. And finally, according to Darcy’s
law, viscous forces are associated to the pressure gradients near the well bore. Normally
there is a balance in any point of the reservoir between gravitational and viscous forces, but
when viscous forces at the well bore exceed the gravitational forces, a cone is formed that
will eventually break into the well, if not controlled. Under steady state condition pressure
gradients are constant: viscous forces are lower than gravity forces, thus allowing the
formation of a stabilized water cone. Under unsteady state conditions, the cone is formed
and grows up towards the perforated interval until water breaks into the well bore.
According to the definition of Hamed (2000), the critical production rate is the rate above
which the flowing pressure gradient at the well causes water to cone breaking into the well.
Therefore, it is the maximum rate of oil production without concurrent production of the
displacing phase by coning. At the critical rate, the built-up cone is stable, but is at a position
of incipient breakthrough. In field operations typical problems are the prediction of the
maximum flow rate that can be produced from a completed well without simultaneous
production of water and, secondly, the definition of the optimum length and position of
intervals to be perforated in a new well in order to obtain the maximum water-free production.
Since the early days of reservoir engineering several experiments and mathematical
analyses were conducted with the aim of understanding coning problems. The models
developed to predict the critical rate can be grouped into two broad categories. They are
either derived from different, at times conflicting, theoretical assumptions or founded on
correlations, and often lead to very different results. Mathematical models are based on the
equilibrium condition of viscous forces and gravity forces by calculation of an oil potential
function, which is solved for the critical rate by allowing viscous forces to equal gravitational
forces. Empirical correlations were developed from laboratory experiments or numerical
simulations. During the last decades the interest of critical rate studies shifted from
experimental models to numerical models, both due to the complexity of reservoir
engineering which make laboratory models impractical and to the recent advances of
computers.
Muskat and Wyckoff (1935) presented an approximate solution of the water–coning problem.
For an isotropic reservoir, the critical rate can be estimated by a graphic procedure. Their
solution is based on three assumptions: the oil potential distribution around the well under
steady-state conditions is given by the solution of Laplace’s equation for incompressible
fluids; a uniform-flux boundary condition is defined at the well bore, giving a varying well
potential with depth; the effect of the cone shape on the potential distribution of the oil phase
is negligible. In Muskay’s study the oil is flowing in a sand formation between two parallel
impermeable boundaries and into a well partially penetrating the formation. Because the
solution is restricted to the conditions considered by the authors, this method was discarded
in this study.
Meyer and Garder (1954) proposed an approximate equation that, assuming radial flow, can
calculate the maximum theoretical oil flow rate with no water production while a water cone
exists. The maximum theoretical flow rate is defined as the production rate at which the
stable cone reaches the bottom of the well perforated interval.
Using Muskat and Wyckoff’s theory and a potentiometric model, in 1964 Chierici and Ciucci
presented a new approach for determining the maximum oil rate without water production.
The critical oil rate can be evaluated through a set of dimensionless graphs that take into
account the reservoir and fluid properties, including the formation anisotropy. This method
also has a limit of validity because it only applies to restricted ranges of the well-reservoir
system geometrical characteristics.
Wheatley (1985) presented an approximate theory for water coning of incompressible fluids
in a stable cone situation. He postulated a potential function containing a linear combination
of line and point sources with three adjustable parameters. The function satisfies Laplace’s
equation, and by properly adjusting the model parameters, Wheatley was able to satisfy the
boundary conditions closely, including that of constant well potential. His theory was the first
to account for the cone shape by requiring the cone surface to be a streamline. Included in

2
the paper is a fairly simple procedure for predicting critical rate as a function of
dimensionless well radius and well penetration for anisotropic formations.
Chaperon (1986) compared the critical rates in vertical and horizontal wells, studying the
behaviour of cresting toward horizontal wells in an anisotropic formation assuming a constant
oil-water interface elevation at a finite distance from the well. Since the author neglected the
flow restriction due to the immobile water in the crest, the theory might give an over optimistic
evaluation of the critical rate.
Abass and Bass (1988) studied the phenomenon of water coning under different boundary
conditions analytically, numerically, and experimentally. In their study, a fully implicit, strongly
coupled mathematical model was used to handle rapid pressure and saturation changes. In
addition, a plexiglas model was constructed to obtain qualitative and quantitative description
of water-coning. In their experimental study no critical oil rate for an unstable cone was
observed. They derived an analytical solution for calculating the water-free oil rate for steady
state and pseudo-steady state flow conditions in a 2-D radial flow system using an average
pressure concept. Although the adoption of their 2-D radial flow assumption and average
pressure concept may not be suitable for water-coning systems, they were the first
investigators who took into account the effect of limited well bore penetration on the
maximum available water-free oil rate. According to their solution, the optimum fractional well
bore penetration interval (completed from the top of the formation) should be 0.5 in oil-water
coning systems.
Joshi (1988, 1991) studied the productivity of slanted and horizontal wells using the potential
fluid theory. The mathematical solution of his theory was simplified by subdividing the three-
dimensional fluid flow problem into two bi-dimensional problems, namely oil flow into a
horizontal well in a horizontal plane and in a vertical plane. He also compared horizontal
wells to vertical wells in terms of critical rate and gas and water coning occurrence, providing
an equation for the calculation of critical rates for water coning systems.
Høyland et al. (1989) presented two methods, analytical and numerical respectively, to
predict critical oil rates in anisotropic formations produced by a well completed from the top
of the reservoir. The analytical solution is similar to that of Muskat and Wickoff (1935) except
that the assumption of uniform flux at the well bore was replaced by the assumption of an
infinite conductive well bore. The numerical method was based on a calculation made with a
simulation model. The aquifer was represented by a bottom layer with infinite porosity and a
column of infinite porosity at the drainage radius.
Giger (1989) presented an analytical 2-D mathematical model of water cresting before
breakthrough for horizontal wells. Since he used the free surface boundary condition and
assumed that the free surface is at a large distance from the well, the oil height in the model
may be difficult to choose. Although the author modified the mathematical solution based on
the comparison with experimental data, he suggested that these solutions should not be
used for small values of the drainage radius.
Aulie et al. (1995) investigated both analytically and experimentally the cresting behaviour
towards horizontal wells with bottom water drive and edge water drive based on the work of
Giger and using laboratory models. For bottom water drive the experimental results were
compared with two models for transient crestal behaviour. He assumed pressure drawdown
due to linear flow, static vertical equilibrium and static pressure in the water zone. According
to the author reasonable predictions can be obtained for bottom water drive at low flow rates.
In the above-mentioned investigations the effect of well partial penetration on oil productivity
was not taken into account. Therefore, the maximum water-free production rate would occur
when the well bore penetration in the reservoir is zero, which is physically impossible.
Furthermore, Guo and Lee (1993) presented an analytical solution to calculate the critical oil
rate and the optimum completion interval for a well that partially penetrates an isotropic
formation. They assumed a radial-spherical combined flow instead of a radial bi-dimensional
flow. The optimum well bore penetration into an oil zone was determined theoretically. The
analytical solution showed that the optimum penetration into a pay zone was less than one-
third the total reservoir thickness.

3
Although many researchers have studied the coning behaviour to determine the critical oil
rate, their results frequently conflict because they made different assumptions to simplify the
problem. In addition, while the existence of a critical oil rate associated with an unstable cone
was proved analytically (Muskat and Wyckoff, 1935; Wheatley, 1985) and was detected
experimentally (Aulie et al., 1995), it was not always observed (Abass and Bass, 1988).
It should be noted that neither analytical methods nor experimental methods have been
checked against field data the same way numerical methods have been. Menouar and
Hakim (1993) conducted a study of water coning and cresting testing a numerical model
against field data and adopting a method based on the observation of the variation of the
saturation gradient in the reservoir. They found that their method compared favourably to the
most recent correlations.
As several scientists believe, in principle analytical models should be preferred over
numerical simulations for the ease of their applicability. However, they often fail to reliably
describe real systems because of their inherent limitations, and numerical models represent
the only viable option for complex reservoir problems. The objective of the present study was
to conduct an exhaustive comparison among the most used analytical methods for water
coning and water cresting prediction and the results obtained by numerical simulation for a
number of possible scenarios characterised by different combination of petrophysical and
fluid properties. The main goal was to verify whether analytical models could be used as
alternative to numerical models, taken as a reference. In fact, provided that the reservoir is
reasonably well characterised, although time-consuming, numerical models are indeed a
reliable tool for describing any type of fluid flow problem. The results provided by the
analytical models are compared to numerical simulations and discussed with respect to their
simplifying assumptions. Eventually, field production data of a vertical well and of a horizontal
well were also considered for comparison.

METHODOLOGY

The most used analytical methods for predicting the critical oil rate were selected, namely the
models proposed by Meyer and Garder, Chaperon, Chierici and Ciucci, Wheatley, Høyland
et al., Abass and Bass, Guo and Lee for vertical wells and Chaperon, Giger, and Joshi for
horizontal wells. Each method was analysed paying attention to the main assumptions and
the validity range of its solutions (Cuda, 2006).
The results provided by the analytical methods were compared against the results obtained
by numerical simulations. Several preliminary tests were performed to select the numerical
models most suited for this study. A Cartesian and a radial model geometry were considered
for simulating water coning under vertical wells. The radial geometry allowed reaching the
best compromise between computational time and accuracy of results. In the case of
horizontal wells, the results of a cross section model were compared to the results obtained
with a full 3D model, and good accordance between the two models was found. The cross
section was selected due to the reduced computational time it required. Therefore, two
different models were then adopted, a radial model for studying water coning and a bi-
dimensional cross section for describing water cresting. The radial model is shown in fig.1,
and is characterised by 21 cells in the horizontal direction (cell dimensions increase with a
logarithmic progression) and 30 cells in the vertical direction; each cell is 1-meter thick. The
Cartesian model is shown in fig. 2, and is characterised by 51 cells in the horizontal direction
and 20 cells in the vertical direction; cell dimensions are 5x1.5 meters.
Both numerical models used in the study were two-phase, black oil models with finite-
difference formulation. The original oil/water contact was set at the very bottom of the
reservoir; the water zone was represented by the bottom layer, at which infinite porosity and
permeability were assigned to simulate a constant-pressure boundary. The top and the
bottom of the reservoir were set as no-flow boundaries. The grid orientation and the spatial
discretization were selected with the aim of accurately simulating the water coning or cresting
formation, thus high resolution models were adopted to avoid numerical instability problems.

4
Several simulations were run to verify that gridding effects did not affect the solution, trying to
achieve an approximately constant potential drop between grid blocks at steady-state
conditions.

Fig 1: Section of the radial model adopted to evaluate the critical oil rate for vertical
wells with initial fluid distribution

Fig 2: Bi-dimensional cross-section model adopted to evaluate the critical oil rate for
horizontal wells with initial fluid distribution

The technique adopted for calculating the critical oil rate was based on the adjustment of the
liquid production rate while observing the variation of the water cut, i.e., the ratio between the
water rate and the total liquid rate. For each well type the study was conducted imposing a
constant liquid production for a certain period of time in order to induce steady state
conditions; subsequently, a reduction to the liquid rate was applied to respect almost zero
water production. This procedure allowed the formation and stabilisation of the water cone or
cresting immediately below the perforated intervals. Because the limit applied to water cut
strongly affected both the rapidity with which the critical rate was found and the precision with
which the critical rate was maintained, different values of liquid production rates and water
cut limits were adopted. Moreover, the imposed liquid rate was assumed to be constant
along the horizontal well. The duration of the simulation had to be sufficiently long to find a
satisfying stabilisation of the liquid production rate. Time steps length varied from a minimum
of one day to two month to find a suitable compromise between numerical stability and
computational time.
All the formation and fluid properties were preliminary tested in order to assess which ones
most affected water coning and water cresting. Parameters such as relative permeability
curves, water viscosity, well bore radius and model extent had negligible effects on coning
and cresting phenomena. Absolute permeability, viscosity and formation volume factor, water
and oil density, total reservoir thickness, vertical well penetration, and anisotropy ratio

5
affected the formation of water coning and cresting. These preliminary simulations proved
that water coning is a local phenomenon that depends on the formation and fluid properties
in the well drainage area.
After assessing which parameters affected water coning and cresting problems, a base case
scenario was selected, represented by a light oil, homogeneous and isotropic reservoir. A
number of other scenarios were also considered in the study for a thorough comparison
between numerical models and analytical models. The model properties that were not
changed during the simulations are summarized in tab. 1. Tab. 2 provides the values of the
model parameters assigned to the reference case and the range within which they were
varied during the study. Each parameter was varied independently.

Tab. 1: Reference case scenario

Parameter Value Parameter Value

External radius (m) Rock compressibility


500 5·10-5
(vertical well) (1/barsa)
External radius (m)
125 Water viscosity (cP) 1
(horizontal well)
Well radius (m) 0.15 Oil density (kg/m3) 876
Water compressibility
Reservoir thickness (m) 30 1·10-5
(1/barsa)
Porosity (-) 0.2 Water density (kg/m3) 1060

Both numerical and analytical models were then applied to field data to further test their
ability to predict critical oil rates. A vertical well and a horizontal well that were produced
water-free for two years were considered. Real fluid and rock properties were used in the
models and their prediction of the critical oil rates compared to the historical data. Due to the
local nature of the coning/cresting phenomena no full field model was built for this phase of
the study, and the same simplified numerical models as for all the other simulations were
used.

Tab. 2: Reference case scenario and parameter variations considered in the


sensitivity analyses

Reference case Range of values in


Parameter
value simulations
Permeability (mD) 100 50 - 1000

Anisotropy ratio (-) 1 0.1 - 1

Oil viscosity (cP) 1 0.5 - 50

Partial penetration rate (-) 0.2 0.2 - 0.8


Distance between oil-water contact and
20 10 - 25
horizontal well (m)

RESULTS AND DISCUSSION

The results of the sensitivity analysis are presented in terms of critical oil rate as a function of
absolute permeability (homogeneous isotropic case), anisotropy ratio, oil viscosity, fractional
well penetration (for the vertical well case only), and distance between the oil-water contact

6
and the well (for the horizontal well case only), for both analytical and numerical models. The
response of the numerical models at the stabilisation of the water coning or cresting (fig. 3
and fig. 4, respectively) was taken as a reference, so that the results obtained with the
analytical methods were evaluated with respect to the numerical simulation results. The
results obtained with the method proposed by Guo and Lee are not represented in the
graphs because they severely overestimated the oil rate for all scenarios due to the
assumption of radial and spherical composite flow, as confirmed by the same authors in their
study.

Fig 3: Water coning formation for vertical wells

Fig 4: Water cresting formation for horizontal wells

The trend of the critical oil rate for vertical wells as a function of absolute permeability in
homogeneous isotropic reservoirs is shown in fig. 5. The response of the analytical models is
linear, and generally provides an underestimation of the critical oil rate. The results given by
Chaperon and Chierici and Ciucci are in good agreement because they do not consider the
presence of the cone on the potential distribution and they both calculated the critical oil rate
at the external radius, where the oil-water contact is undisturbed.
Although the Wheatley model provides critical rates which are very close to the values
obtained from numerical simulations, especially for high permeability values, the method is
unpractical to apply because the calculation of the critical oil rate implies iterative numerical
procedures.

7
120

Meyer
100
Chierici

critical oil rate [m /day]


Chaperon
Hoyland

3
80
Abass
Wheatley
60 Numerical Model

40

20

0
0 100 200 300 400 500 600 700 800 900 1000
k [mD]

Fig 5: Vertical wells - critical oil rates as a function of absolute permeability


(homogeneous isotropic reservoir)

The models proposed by Meyer and Garder and by Abass and Bass strongly underestimate
the critical oil rates, due to restrictive hypotheses adopted in calculation method. Meyer and
Garder’s model underestimates the critical oil rate because the production rate is evaluated
when a stable cone reaches the perforated interval, thus inducing the oil to flow through a
section comprised between the top of the formation and the apex of the cone. The main
limitations of Abass and Bass’ model are the adoption of the average pressure in the
calculation of the critical rate and the assumption of bi-dimensional radial flow, which is not
suitable to describe local water coning phenomena.
Fig. 6 shows the critical oil rate as a function of the anisotropy ratio. Meyer and Garder’s
model does not account for formation anisotropy, thus the critical oil rate keeps constant. The
critical oil rate generally decreases for increasing values of the anisotropy ratio, with the
exception of Abass and Bass’ model where the anisotropy ratio is approximated to the ratio

30
Meyer
Chierici
25 Chaperon
critical oil rate [m /day]

Hoyland
Abass
Wheatley
3

20
Numerical Model

15

10

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
kv/kh [-]

Fig 6: Vertical wells - critical oil rates as a function of the anisotropy ratio

between the external radius of the water cone, which can be properly adjusted, and the oil
column thickness. The response of Guo ad Lee’s model was comparable to the other
analytical methods for low anisotropy ratios, when the vertical components of the spherical

8
flow become negligible and the prevailing flow is radial. Again, this proved the inadequacy of
the radial-spherical composite flow to represent the physics of water coning systems.
Fig. 7 shows the critical oil rates as a function of oil viscosity.

30

Meyer
25 Chierici
critical oil rate [m /day] Chaperon
Hoyland
3

20
Abass
Wheatley
15 Numerical Model

10

0
0.1 1 10 100
μo [cp]

Fig 7: Vertical wells - critical oil rates as a function of oil viscosity

The critical oil rate decreases for increasing values of oil viscosity. Although both the
analytical methods and the numerical simulation show the same trend, once again the critical
oil rates were underestimated by all the analytical models, especially for low oil viscosity
values.
The critical oil rates calculated as a function of the fractional well penetration are presented
in fig. 8. Three different trends were identified.

20

18 Meyer
Chierici
critical oil rate [m /day]

16 Chaperon
14 Hoyland
3

Abass
12 Wheatley
10 Numerical Model

0
0 0.2 0.4 0.6 0.8 1
Fractional well penetration [-]

Fig 8: Vertical wells - critical oil rates as a function of the fractional well penetration

Chaperon’s model is insensitive to the well penetration in the reservoir. The methods
proposed by Chierici and Ciucci and by Høyland et al. are applicable only for limited ranges
of external radius and well penetration, while Wheatley’s and Meyer and Garder’s models do
not consider the well penetration (theoretically, it is zero). The only two methods that
respected the trend of the results given by numerical simulations are Abass’ and Guo and
Lee’s. According to Abass and Bass, a fractional well penetration of 0.5-0.6 would lead to the

9
maximum critical oil rates, while Guo and Lee affirm that the optimal range is 0.3-0.4.
Numerical simulations showed that the optimal well penetration ranges between 0.1 and 0.2
of the reservoir thickness.
The analytical methods considered to evaluate the critical oil rate for horizontal wells are
different from those examined in the case of vertical wells with the exception of Chaperon’s
method, which can also describe water cresting phenomena. Fig. 9 shows the critical oil
rates as a function of absolute permeability in homogeneous, isotropic reservoirs.

450

400
Chaperon
critical oil rate [m /day]

350 Giger
Joshi
3

300
Numerical model
250

200

150

100

50

0
0 100 200 300 400 500 600 700 800 900 1000

kh [mD]

Fig 9: Horizontal wells - critical oil rates as a function of absolute permeability


(homogeneous isotropic reservoirs)

In general, the analytical methods by Chaperon and Giger overestimate the critical oil rate
with respect to numerical simulations. Chaperon’s model provides the highest critical oil rates
in every scenario, because it does not account for the presence of the water coning in the
formulation of the potential equation. Moreover, according to the model the drain is located at
the top of the reservoir. Joshi’s model appears to be in very good agreement with the
numerical model.
The critical oil rates as a function of anisotropy ratio are represented in fig. 10.
70

Chaperon
60 Giger
critical oil rate [m /day]

Joshi
50 Numerical Model
3

40

30

20

10

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

kV/kH [-]

Fig 10: Horizontal wells - critical oil rates as a function of the anisotropy ratio

A part from Giger’s model which is only applicable to homogeneous reservoirs, all the other
models seem to be insensitive to variations of the anisotropy ratio. As discussed before,
10
Chaperon’s method largely overestimates the results provided by numerical simulations due
to the absence of the water cresting in the potential distribution in the reservoir.
Fig. 11 shows the critical oil rates as a function of oil viscosity.

90

80

critical oil rate [m /day] 70 Chaperon


Giger
3

60
Joshi
Numerical Model
50

40

30

20

10

0
0.1 1 10 100

μo [cp]

Fig 11: Horizontal wells - critical oil rates as a function of oil viscosity

As for vertical wells, an inverse relation between critical oil rates and oil viscosity was found.
Although the analytical models reproduce the same trend as the numerical simulations, they
overestimated the critical rates except for Joshi’s model, which is in excellent agreement with
the numerical model. Again, the error in predicting the critical oil rate is higher for low oil
viscosity values.
Fig. 12 shows the critical oil rates as a function of the distance between the oil-water contact
and the horizontal well.

13

12
critical oil rate [m /day]

11
3

10

7
Joshi
6 Numerical Model

4
5 10 15 20 25 30

distance between WOC and horizontal well [m]

Fig 12: Horizontal wells - critical oil rates as a function of the distance between the oil-
water contact and the horizontal well

Among the considered analytical models, only Joshi’s method accounts for the position of the
horizontal drain in the reservoir. The calculated critical oil rates are slightly overestimated
compared with those obtained by numerical simulations.
Based on the obtained results, the general conclusion can be drawn that analytical models
are not reliable for predicting critical oil rates, except for Joshi’s model. Critical oil rates

11
values as calculated with Joshi’s model are always in very good agreement with those
obtained from numerical simulations.
In order to prove that the results obtained from numerical simulations were representative of
real systems, both the radial and the cross-section models were validated on real production
data. For this purpose, a vertical and a horizontal wells were selected among 20 wells
belonging to a real field. The reservoir is a heavy oil, highly heterogeneous formation and
has been produced for over 30 years. The geometrical parameters of the vertical well system
and those of the horizontal well system are shown in fig 13. The selected wells have been
producing at the critical rate during the last two years.

horizontal well length = 300 m

(a) (b)

Fig 13: Geometrical parameters of the systems representing the real cases: (a) vertical
well and (b) horizontal well

The analytical and numerical models were used to calculate the critical oil rates based on the
real reservoir and well characteristics. The adopted permeability values correspond to the
average permeability of the drainage area of each well based on the history-matched full field
model (k = 100 mD for the case of the vertical well and k = 80 mD for the case of the
horizontal well).
Errors on critical oil rates with respect to the historical production data of the two wells are
summarized in tab. 3. Numerical simulations could predict the critical oil rate better than any
of the analytical methods for both the vertical and the horizontal well cases. In general, errors
found in the sensitivity analyses were confirmed. The application of the model by Chierici and
Ciucci was impossible because the geometrical parameters of the real case were out of the
range of validity of the method; in addition the algorithm of Wheatley’s method was unstable.
In the case of the horizontal well, despite the fact that a very good agreement with the results
of the numerical simulations had been found in the sensitivity analyses, Joshi’s model
underestimated the real critical oil rate thus proving unreliable.

12
Tab. 3: Errors on critical rates as calculated with analytical and numerical models
with respect to the historical production data for the validation cases

Vertical well Horizontal well


(critical oil rate = 11.5 m3ST/day) (critical oil rate = 2.9 m3ST/day)
Model Error (%) Model Error (%)

Numerical model +4.2 Numerical model -0.6

Meyer and Garder -87.2 Giger +250

Chaperon -40 Joshi -33

Chierici and Ciucci - Chaperon +377

Wheatley -

Høyland et al. -53

Abass and Bass -91

Guo and Lee +4200

CONCLUSIONS

Water coning and water cresting phenomena may affect the efficiency of oil drainage from
the reservoir, thus reducing the ultimate oil recovery, increase the cost of production
operations, and arise environmental problems related to water disposal. The production of oil
may be uneconomical if the water phase becomes the dominant one, as a consequence the
well must be shut in. The critical oil rate is the rate at which water free oil is produced.
Because a rigorous approach of general validity could not be found in the literature, the
objective of this study was to provide a comprehensive comparison between the most used
analytical methods and the results obtained by numerical models, for a number of scenarios
and for vertical and horizontal wells. Two different numerical models were adopted for the
simulations, and their response was assumed to be representative of the real reservoir
behavior. This assumption was later verified validating the results provided by the analytical
and by the numerical models by comparison against some real data.
Preliminary simulations with the numerical models proved that water coning and cresting are
a local phenomena that depend on the formation and fluid properties in the well drainage
area.
In general, the obtained results proved that analytical models represent a fast but imprecise
tool for predicting critical oil rates. If qualititatively compared to numerical simulations,
analytical models correctly reproduce the influence of the parameters that mostly affect water
coning and cresting. However, from a quantitative point of view, the results given by
analytical models were rarely comparable with those obtained with numerical simulations.
Analytical models typically underestimate the critical oil rate by more than 30% with respect
of numerical models. The models by Guo and Lee and by Abass and Bass seem to be the
most complete models, but they strongly overestimates or underestimates the critical oil rate.
The research showed the inadequacy of analytical models for the evaluation of critical oil
rates in the case of horizontal wells. It was verified that in general they largely overestimate
the critical oil rate. The model proposed by Joshi proved highly reliable in the sensitivity
analyses, but failed to reproduce the critical oil rate measured in the real case.
The recommendation is made to use of analytical models only as a first approximation for a
qualitative estimation of oil critical rates. A reliable prediction of the critical oil rate can only
be achieved by high resolution numerical models.

13
ACKNOWLEDGEMENTS

The Politecnico di Torino greatly acknowledges Schlumberger GeoQuest for the software
used in this research.
The authors are very grateful to Carlo Bruni for his valuable contribution to numerical
modelling.

REFERENCES

Abass H. H., Bass D. M., “The Critical Production Rate in Water-Coning System”, Paper SPE
17311 presented at the SPE Permian Basin Oil and Gas Recovery Conference, Midland
(TX), 10-11 March 1988.
Aulie T., Grødal E., Asheim H., Oudemann P., “Experimental Investigation of Cresting and
Critical Flow Rate of Horizontal Wells“. Paper SPE 26639. SPE Advanced Technology
Series, Vol. 3, No. 1, 1995.
Chaperon I., “Theoretical Study of Coning Toward Horizontal and Vertical Wells in
Anisotropic Formations: Subcritical and Critical Rates”, Paper SPE 15377 presented at the
61st SPE Annual Technical Conference and Exhibition, New Orleans (LA), 5-8 October 1986.
Chierici G.L., Ciucci G. M., Pizzi G., “A Systematic Study of Gas and Water Coning By
Potentiometric Models”, JPT, August 1964. 923-29; Trans. AIME, 231.
Cuda S., “Evaluation of the critical oil rate with analytical and numerical models”, MSc Thesis
(in Italian), Engineering Faculty, Politecnico di Torino, May 2006.
Giger F. M., “Analytic 2-D Models Of Water Cresting Before Breakthrough for Horizontal
Wells”, Paper SPE 15378 presented at the 61st SPE Annual Technical Conference and
Exhibition, New Orleans (LA), 5-8 October 1986.
Guo B., Lee R.L.-H., “A Simple Approach to Optimisation of Completion Interval in Oil/Water
Coning System”, SPERE Vol. 8 No. 4, November 1993, pp. 249-255.
Hamed T., "Reservoir engineering handbook", Gulf Publishing Company, 2000.
Høyland L. A., Papatzacos P., Skjaeveland S.M., “Critical Rate for Water Coning; Correlation
and Analytical Solution”, SPERE Vol. 4 No. 4, November 1989, pp. 495-502.
Joshi S.D., “Augmentation of Well Productivity Using Slant and Horizontal Wells”, JPT No.6,
June 1988, pp. 729-739.
Joshi S.D., “Horizontal well technology”, Pennwell Publishing Company, 1991, pp. 535.
Menouar H.K., Hakim A. A., “Water Coning and Critical Rates in Vertical and Horizontal
Wells”, Paper SPE 29877 presented at he SPE Middle East Oil Show, Bahrain, 11-14 March
1995.
Meyer H.L., Garder A.O., “Mechanics of Two Immiscible Fluids in Porous Media”, Journal of
Applied Physics, Vol. 25, No. 11, 1954.
Muskat, M., Wyckoff R. D., “An Approximate Theory of Water Coning in Oil Production”,
Trans. AIME, Vol. 114, 1935.
Wheatley M.J., “An Approximate Theory of Oil/Water Coning”, Paper SPE 14210 presented
at the 60th SPE Annual Technical Conference and Exhibition, Las Vegas (NV), 22-25
September 1985.

14

You might also like